This PDF file contains the front matter associated with SPIE Proceedings Volume 9783, including the Title Page, Copyright information, Table of Contents, and Conference Committee listing.

A conventional 3D/4D digital subtraction angiogram (DSA) requires two rotational acquisitions (mask and fill) to compute the log-subtracted projections that are used to reconstruct a 3D/4D volume. Since all of the vascular information is contained in the fill acquisition, it is hypothesized that it is possible to reduce the x-ray dose of the mask acquisition substantially and still obtain subtracted projections adequate to reconstruct a 3D/4D volume with noise level comparable to a full dose acquisition. A full dose mask and fill acquisition were acquired from a clinical study to provide a known full dose reference reconstruction. Gaussian noise was added to the mask acquisition to simulate a mask acquisition acquired at 10% relative dose. Noise in the low-dose mask projections was reduced with a weighted edge preserving (WEP) filter designed to preserve bony edges while suppressing noise. 2D log-subtracted projections were computed from the filtered low-dose mask and full-dose fill projections, and then 3D/4D-DSA reconstruction algorithms were applied. Additional bilateral filtering was applied to the 3D volumes. The signal-to-noise ratio measured in the filtered 3D/4D-DSA volumes was compared to the full dose case. The average ratio of filtered low-dose SNR to full-dose SNR was 1.07 for the 3D-DSA and 1.05 for the 4D-DSA, indicating the method is a feasible approach to restoring SNR in DSA scans acquired with a low-dose mask. The method was also tested in a phantom study with full dose fill and 22% dose mask.

Multiplanar reconstruction (MPR) in digital breast tomosynthesis (DBT) allows tomographic images to be portrayed in various orientations. We have conducted research to determine the resolution of tomosynthesis MPR. We built a phantom that houses a star test pattern to measure resolution. This phantom provides three rotational degrees of freedom. The design consists of two hemispheres with longitudinal and latitudinal grooves that reference angular increments. When joined together, the hemispheres form a dome that sits inside a cylindrical encasement. The cylindrical encasement contains reference notches to match the longitudinal and latitudinal grooves that guide the phantom’s rotations. With this design, any orientation of the star-pattern can be analyzed. Images of the star-pattern were acquired using a DBT mammography system at the Hospital of the University of Pennsylvania. Images taken were reconstructed and analyzed by two different methods. First, the maximum visible frequency (in line pairs per millimeter) of the star test pattern was measured. Then, the contrast was calculated at a fixed spatial frequency. These analyses confirm that resolution decreases with tilt relative to the breast support. They also confirm that resolution in tomosynthesis MPR is dependent on object orientation. Current results verify that the existence of super-resolution depends on the orientation of the frequency; the direction parallel to x-ray tube motion shows super-resolution. In conclusion, this study demonstrates that the direction of the spatial frequency relative to the motion of the x-ray tube is a determinant of resolution in MPR for DBT.

We have developed a clinically ready first generation stationary breast tomosynthesis system (s-DBT). In the s-DBT system, focal spot blur associated with x-ray source motion is completely eliminated, allowing for rapid acquisition of projection images over a larger angular span without changing the acquisition time.

In the phantom studies the 1st generation s‐DBT system has demonstrated 30% higher spatial resolution than the corresponding continuous motion DBT systems. The system is currently being evaluated for its diagnostic performance in 100 patient clinical evaluation against FFDM. Initial results indicate that the s­‐DBT system can produce increased lesion conspicuity and comparable MC visibility. However due to x­‐ray flux limitations, certain large size patients have to be excluded. Recent studies have shown that increasing the angular span beyond 30° can be beneficial for enhanced depth resolution. We report the preliminary characterization of the 2nd generation s-­DBT system with a new CNT x-­ray source array, increased tube flux and a larger angular span. Increasing x‐ray tube flux allows for a larger patient population and dual energy imaging. Results indicate that the system delivers more than twice the flux, allowing for imaging of all size patients with acquisition time of 2­‐4 seconds. A 7° increase in angular span over 1st generation decreased the ASF by 37%. Additionally, the 2nd generation s‐DBT system utilizing a specific AFVR reconstruction method resulted in a 92% increase in the in plane resolution over CM DBT system, and a 37% increase in spatial resolution over the 1st generation s-‐DBT system.

The coronary artery calcium score (CACS) measures the buildup of calcium on the coronary artery wall and has been shown to be an important predictor of the risk of coronary artery diseases (CAD). Currently CACS is measured using CT, though the relatively high cost and high radiation dose has limited its adoption as a routine screening procedure. Digital Chest Tomosynthesis (DCT), a low dose and low cost alternative to CT, and has been shown to achieve 90% of sensitivity of CT in lung disease screening. However commercial DCT requires long scanning time and cannot be adapted for high resolution gated cardiac imaging, necessary for CACS. The stationary DCT system (s- DCT), developed in our lab, has the potential to significantly shorten the scanning time and enables high resolution cardiac gated imaging. Here we report the preliminary results of using s-DCT to estimate the CACS. A phantom heart model was developed and scanned by the s-DCT system and a clinical CT in a phantom model with realistic coronary calcifications. The adapted fan-beam volume reconstruction (AFVR) method, developed specifically for stationary tomosynthesis systems, is used to obtain high resolution tomosynthesis images. A trained cardiologist segmented out the calcifications and the CACS was obtained. We observed a strong correlation between the tomosynthesis derived CACS and CT CACS (r² = 0.88). Our results shows s-DCT imaging has the potential to estimate CACS, thus providing a possible low cost and low dose imaging protocol for screening and monitoring CAD.

The aim of this study was to compare the detection of microcalcification clusters by human observers in breast images using 2D-mammography and narrow (15°/15 projections) and wide (50°/25 projections) angle digital breast tomosynthesis (DBT). Simulated microcalcification clusters with a range of microcalcification diameters (125 μm-275 μm) were inserted into 6 cm thick simulated compressed breasts. Breast images were produced with and without inserted microcalcification clusters using a set of image modelling tools, which were developed to represent clinical imaging by mammography and tomosynthesis. Commercially available software was used for image processing and image reconstruction. The images were then used in a series of 4-alternative forced choice (4AFC) human observer experiments conducted for signal detection with the microcalcification clusters as targets. The minimum detectable calcification diameter was found for each imaging modality: (i) 2D-mammography: 164±5 μm (ii) narrow angle DBT: 210±5 μm, (iii) wide angle DBT: 255±4 μm. A statistically significant difference was found between the minimum detectable calcification diameters that can be detected by the three imaging modalities. Furthermore, it was found that there was not a statistically significant difference between the results of the five observers that participated in this study. In conclusion, this study presents a method that quantifies the threshold diameter required for microcalcification detection, using high resolution, realistic images with observers, for the comparison of DBT geometries with 2D-mammography. 2Dmammography can visualise smaller detail diameter than both DBT imaging modalities and narrow-angle DBT can visualise a smaller detail diameter than wide-angle DBT.

Dual-energy contrast-enhanced digital mammography (DE CE-DM) uses an iodinated contrast agent to image the perfusion and vasculature of the breast. DE images are obtained by a weighted logarithmic subtraction of the high-energy (HE) and low-energy (LE) image pairs. We hypothesized that the optimal DE subtraction weighting factor is thickness-dependent, and developed a method for determining breast tissue composition and thickness in DE CE-DM. Phantoms were constructed using uniform blocks of 100% glandular-equivalent and 100% adipose-equivalent material. The thickness of the phantoms ranged from 3 to 8 cm, in 1 cm increments. For a given thickness, the glandular-adipose composition of the phantom was varied using different combinations of blocks. The logarithmic LE and logarithmic HE signal intensities were measured; they decrease linearly with increasing glandularity for a given thickness. The signals decrease with increasing phantom thickness and the x-ray signal decreases linearly with thickness for a given glandularity. As the thickness increases, the attenuation difference per additional glandular block decreases, indicating beam hardening. From the calibration mapping, we have demonstrated that we can predict percent glandular tissue and thickness when given two distinct signal intensities. Our results facilitate the subtraction of tissue at the boundaries of the breast, and aid in discriminating between contrast agent uptake in glandular tissue and subtraction artifacts.

Virtual clinical trials are an emergent approach for the rapid evaluation and comparison of various breast imaging technologies and techniques using computer-based modeling tools. A fundamental requirement of this approach for mammography is the use of realistic looking breast anatomy in the studies to produce clinically relevant results. In this work, a biologically inspired approach has been used to simulate realistic synthetic breast phantom blocks for use in virtual clinical trials. A variety of high and low frequency features (including Cooper’s ligaments, blood vessels and glandular tissue) have been extracted from clinical digital breast tomosynthesis images and used to simulate synthetic breast blocks. The appearance of the phantom blocks was validated by presenting a selection of simulated 2D and DBT images interleaved with real images to a team of experienced readers for rating using an ROC paradigm. The average areas under the curve for 2D and DBT images were 0.53±.04 and 0.55±.07 respectively; errors are the standard errors of the mean. The values indicate that the observers had difficulty in differentiating the real images from simulated images. The statistical properties of simulated images of the phantom blocks were evaluated by means of power spectrum analysis. The power spectrum curves for real and simulated images closely match and overlap indicating good agreement.

An anthropomorphic digital breast phantom has been developed with the goal of generating random voxelized breast models that capture the anatomic variability observed in vivo. This is a new phantom and is not based on existing digital breast phantoms or segmentation of patient images. It has been designed at the outset to be modality agnostic (i.e., suitable for use in modeling x-ray based imaging systems, magnetic resonance imaging, and potentially other imaging systems) and open source so that users may freely modify the phantom to suit a particular study. In this work we describe the modeling techniques that have been developed, the capabilities and novel features of this phantom, and study simulated images produced from it. Starting from a base quadric, a series of deformations are performed to create a breast with a particular volume and shape. Initial glandular compartments are generated using a Voronoi technique and a ductal tree structure with terminal duct lobular units is grown from the nipple into each compartment. An additional step involving the creation of fat and glandular lobules using a Perlin noise function is performed to create more realistic glandular/fat tissue interfaces and generate a Cooper’s ligament network. A vascular tree is grown from the chest muscle into the breast tissue. Breast compression is performed using a neo-Hookean elasticity model. We show simulated mammographic and T₁-weighted MRI images and study properties of these images.

Spectral imaging is the acquisition of multiple images of an object at different energy spectra. In mammography, dual-energy imaging (spectral imaging with two energy levels) has been investigated for several applications, in particular material decomposition, which allows for quantitative analysis of breast composition and quantitative contrast-enhanced imaging. Material decomposition with dual-energy imaging is based on the assumption that there are two dominant photon interaction effects that determine linear attenuation: the photoelectric effect and Compton scattering. This assumption limits the number of basis materials, i.e. the number of materials that are possible to differentiate between, to two. However, Rayleigh scattering may account for more than 10% of the linear attenuation in the mammography energy range. In this work, we show that a modified version of a scanning multi-slit spectral photon-counting mammography system is able to acquire three images at different spectra and can be used for triple-energy imaging. We further show that triple-energy imaging in combination with the efficient scatter rejection of the system enables measurement of Rayleigh scattering, which adds an additional energy dependency to the linear attenuation and enables material decomposition with three basis materials. Three available basis materials have the potential to improve virtually all applications of spectral imaging.

Mammography is currently the standard imaging modality used to screen women for breast abnormalities and, as a result, it is a tool of great importance for the early detection of breast cancer. Physical phantoms are commonly used as surrogates of breast tissue to evaluate some aspects of the performance of mammography systems. However, most phantoms do not reproduce the anatomic heterogeneity of real breasts. New fabrication technologies, such as 3D printing, have created the opportunity to build more complex, anatomically realistic breast phantoms that could potentially assist in the evaluation of mammography systems. The primary objective of this work is to present a simple, easily reproducible methodology to design and print 3D objects that replicate the attenuation profile observed in real 2D mammograms. The secondary objective is to evaluate the capabilities and limitations of the competing 3D printing technologies, and characterize the x-ray properties of the different materials they use. Printable phantoms can be created using the open-source code introduced in this work, which processes a raw mammography image to estimate the amount of x-ray attenuation at each pixel, and outputs a triangle mesh object that encodes the observed attenuation map. The conversion from the observed pixel gray value to a column of printed material with equivalent attenuation requires certain assumptions and knowledge of multiple imaging system parameters, such as x-ray energy spectrum, source-to-object distance, compressed breast thickness, and average breast material attenuation. A detailed description of the new software, a characterization of the printed materials using x-ray spectroscopy, and an evaluation of the realism of the sample printed phantoms are presented.

Breast ultrasound tomography is an emerging imaging modality to reconstruct the sound speed, density, and ultrasound attenuation of the breast in addition to ultrasound reflection/beamforming images for breast cancer detection and characterization. We recently designed and manufactured a new synthetic-aperture breast ultrasound tomography prototype with two parallel transducer arrays consisting of a total of 768 transducer elements. The transducer arrays are translated vertically to scan the breast in a warm water tank from the chest wall/axillary region to the nipple region to acquire ultrasound transmission and reflection data for whole-breast ultrasound tomography imaging. The distance of these two ultrasound transducer arrays is adjustable for scanning breasts with different sizes. We use our breast ultrasound tomography prototype to acquire phantom and in vivo patient ultrasound data to study its feasibility for breast imaging. We apply our recently developed ultrasound imaging and tomography algorithms to ultrasound data acquired using our breast ultrasound tomography system. Our in vivo patient imaging results demonstrate that our breast ultrasound tomography can detect breast lesions shown on clinical ultrasound and mammographic images.

In this article, we present an easy-to-implement Multi-energy CT scanning strategy and a corresponding reconstruction method, which facilitate spectral CT imaging by improving the data efficiency the number-of-energy- channel fold without introducing visible limited-angle artifacts caused by reducing projection views. Leveraging the structure coherence at different energies, we first pre-reconstruct a prior structure information image using projection data from all energy channels. Then, we perform a k-means clustering on the prior image to generate a sparse dictionary representation for the image, which severs as a structure information constraint. We com- bine this constraint with conventional compressed sensing method and proposed a new model which we referred as Joint Clustering Prior and Sparsity Regularization (CPSR). CPSR is a convex problem and we solve it by Alternating Direction Method of Multipliers (ADMM). We verify our CPSR reconstruction method with a numerical simulation experiment. A dental phantom with complicate structures of teeth and soft tissues is used. X-ray beams from three spectra of different peak energies (120kVp, 90kVp, 60kVp) irradiate the phantom to form tri-energy projections. Projection data covering only 75◦ from each energy spectrum are collected for reconstruction. Independent reconstruction for each energy will cause severe limited-angle artifacts even with the help of compressed sensing approaches. Our CPSR provides us with images free of the limited-angle artifact. All edge details are well preserved in our experimental study.

Compared to conventional computed tomography (CT), dual energy CT allows for improved material decomposition by conducting measurements at two distinct energy spectra. Since radiation exposure is a major concern in clinical CT, there is a need for tools to reduce the noise level in images while preserving diagnostic information. One way to achieve this goal is the application of image-based denoising algorithms after an analytical reconstruction has been performed. We have developed a modified dictionary denoising algorithm for dual energy CT aimed at exploiting the high spatial correlation between between images obtained from different energy spectra. Both the low-and high energy image are partitioned into small patches which are subsequently normalized. Combined patches with improved signal-to-noise ratio are formed by a weighted addition of corresponding normalized patches from both images. Assuming that corresponding low-and high energy image patches are related by a linear transformation, the signal in both patches is added coherently while noise is neglected. Conventional dictionary denoising is then performed on the combined patches. Compared to conventional dictionary denoising and bilateral filtering, our algorithm achieved superior performance in terms of qualitative and quantitative image quality measures. We demonstrate, in simulation studies, that this approach can produce 2d-histograms of the high- and low-energy reconstruction which are characterized by significantly improved material features and separation. Moreover, in comparison to other approaches that attempt denoising without simultaneously using both energy signals, superior similarity to the ground truth can be found with our proposed algorithm.

In this work, we provide an initial characterization of a novel twin robotic X-ray system. This system is equipped with two motor-driven telescopic arms carrying X-ray tube and flat-panel detector, respectively. 2D radiographs and fluoroscopic image sequences can be obtained from different viewing angles. Projection data for 3D cone-beam CT reconstruction can be acquired during simultaneous movement of the arms along dedicated scanning trajectories. We provide an initial evaluation of the 3D image quality based on phantom scans and clinical images. Furthermore, initial evaluation of patient dose is conducted. The results show that the system delivers high image quality for a range of medical applications. In particular, high spatial resolution enables adequate visualization of bone structures. This system allows 3D X-ray scanning of patients in standing and weight-bearing position. It could enable new 2D/3D imaging workflows in musculoskeletal imaging and improve diagnosis of musculoskeletal disorders.

We propose an adapted method of our previously published five-dimensional (5D) motion compensation (MoCo) algorithm1, developed for micro-CT imaging of small animals, to provide for the first time motion artifact-free 5D cone-beam CT (CBCT) images from a conventional flat detector-based CBCT scan of clinical patients. Image quality of retrospectively respiratory- and cardiac-gated volumes from flat detector CBCT scans is deteriorated by severe sparse projection artifacts. These artifacts further complicate motion estimation, as it is required for MoCo image reconstruction. For high quality 5D CBCT images at the same x-ray dose and the same number of projections as todays 3D CBCT we developed a double MoCo approach based on motion vector fields (MVFs) for respiratory and cardiac motion. In a first step our already published four-dimensional (4D) artifact-specific cyclic motion-compensation (acMoCo) approach is applied to compensate for the respiratory patient motion. With this information a cyclic phase-gated deformable heart registration algorithm is applied to the respiratory motion-compensated 4D CBCT data, thus resulting in cardiac MVFs. We apply these MVFs on double-gated images and thereby respiratory and cardiac motion-compensated 5D CBCT images are obtained. Our 5D MoCo approach processing patient data acquired with the TrueBeam 4D CBCT system (Varian Medical Systems). Our double MoCo approach turned out to be very efficient and removed nearly all streak artifacts due to making use of 100% of the projection data for each reconstructed frame. The 5D MoCo patient data show fine details and no motion blurring, even in regions close to the heart where motion is fastest.

Mobile and compact C-arm systems are routinely used in interventional procedures for fluoroscopic CT imaging. The mechanical requirements guarantee for a maximum of flexibility and mobility but restrict the mechanical rotation range (e.g. 165°) and the lateral size of the field of measurement (FOM), typically about 160 mm. Recently, the rotate-plus-shift trajectory for the acquisition of complete datasets from 180° minus fan-angle has been published.^{1, 2} Here, we combine the rotate-plus-shift trajectory with a shifted detector approach for a fully motorized C-arm system. As the isocenter in non-centric C-arms can be freely chosen, the shifted detector can be equally well absorbed with an offset of the C parallel to the transaxial detector direction. The typical rotation range of 360° used in shifted detector trajectories is replaced by a double rotate-plus-shift scan requiring a rotation range of at least 180° minus fan-angle. The trajectory increasing the diameter of the FOM by up to a factor of two is presented and the practical application of variations with an asymmetric FOM is shown. For image reconstruction we use our modified FDK algorithm that is equipped with a generalized redundancy weight. The presented trajectory can increase the applicability and flexibility of C-arm systems and has the potential to perform intra-operative large volume control or overview scans and thus reduce the patient’s risk.

Striped ratio grids are a new concept for scatter management in cone-beam CT. These grids are a modification of conventional anti-scatter grids and consist of stripes which alternate between high grid ratio and low grid ratio. Such a grid is related to existing hardware concepts for scatter estimation such as blocker-based methods or primary modulation, but rather than modulating the primary, the striped ratio grid modulates the scatter. The transitions between adjacent stripes can be used to estimate and subtract the remaining scatter. However, these transitions could be contaminated by variation in the primary radiation. We describe a simple nonlinear image processing algorithm to estimate scatter, and proceed to validate the striped ratio grid on experimental data of a pelvic phantom. The striped ratio grid is emulated by combining data from two scans with different grids. Preliminary results are encouraging and show a significant reduction of scatter artifact.

Purpose: Cone-beam CT (CBCT) of the extremities provides high spatial resolution, but its quantitative accuracy may be challenged by involuntary sub-mm patient motion that cannot be eliminated with simple means of external immobilization. We investigate a two-step iterative motion compensation based on a multi-component metric of image sharpness. Methods: Motion is considered with respect to locally rigid motion within a particular region of interest, and the method supports application to multiple locally rigid regions. Motion is estimated by maximizing a cost function with three components: a gradient metric encouraging image sharpness, an entropy term that favors high contrast and penalizes streaks, and a penalty term encouraging smooth motion. Motion compensation involved initial coarse estimation of gross motion followed by estimation of fine-scale displacements using high resolution reconstructions. The method was evaluated in simulations with synthetic motion (1–4 mm) applied to a wrist volume obtained on a CMOS-based CBCT testbench. Structural similarity index (SSIM) quantified the agreement between motion-compensated and static data. The algorithm was also tested on a motion contaminated patient scan from dedicated extremities CBCT. Results: Excellent correction was achieved for the investigated range of displacements, indicated by good visual agreement with the static data. 10-15% improvement in SSIM was attained for 2-4 mm motions. The compensation was robust against increasing motion (4% decrease in SSIM across the investigated range, compared to 14% with no compensation). Consistent performance was achieved across a range of noise levels. Significant mitigation of artifacts was shown in patient data. Conclusion: The results indicate feasibility of image-based motion correction in extremities CBCT without the need for a priori motion models, external trackers, or fiducials.

C-arm-based cone-beam CT (CBCT) systems with flat-panel detectors are suitable for diagnostic knee imaging due to their potentially flexible selection of CT trajectories and wide volumetric beam coverage. In knee CT imaging, over-exposure artifacts can occur because of limitations in the dynamic range of the flat panel detectors present on most CBCT systems. We developed a straightforward but effective method for correction and detection of over-exposure for an Automatic Exposure Control (AEC)-enabled standard knee scan incorporating a prior low dose scan. The radiation dose associated with the low dose scan was negligible (0.0042mSv, 2.8% increase) which was enabled by partially sampling the projection images considering the geometry of the knees and lowering the dose further to be able to just see the skin-air interface. We combined the line integrals from the AEC and low dose scans after detecting over-exposed regions by comparing the line profiles of the two scans detector row-wise. The combined line integrals were reconstructed into a volumetric image using filtered back projection. We evaluated our method using in vivo human subject knee data. The proposed method effectively corrected and detected over-exposure, and thus recovered the visibility of exterior tissues (e.g., the shape and density of the patella, and the patellar tendon), incorporating a prior low dose scan with a negligible increase in radiation exposure.

For grating interferometer-based x-ray differential phase contrast (DPC) imaging systems, their noise performance is strongly dependent on both the visibility of the interference fringe pattern and the total number of photons used to acquire and extract the DPC signal. For a given interferometer, it is usually designed to work at a specific x-ray energy, therefore any deviation from the designed energy may result in certain visibility loss. In this work, a single photon counting detector (PCD) was incorporated into a DPC imaging system, which enabled photons with energies close to the designed operation energy of the interferometer to be selectively used for DPC signal extraction. This approach led to significant boost in the fringe visibility, but it also discarded x-ray photons with other energies incident on the detector and might result in degradations of the overall radiation dose efficiency of the DPC imaging systems. This work presents a novel singular value decomposition (SVD)-based method to leverage the entire spectrum of x-ray photons detected by the PCD, enabling both fringe visibility improvement and reduction in image noise. As evidenced by the results of experimental phantom studies, the contrast-to-noise ratio of the final DPC images could be effectively improved by the proposed method.

One of the most exciting aspects of the grating based x-ray differential phase contrast (DPC) acquisition method is the concurrent generation of the so-called dark field (DF) signal, along with the classical absorption signal and the novel DPC signal. The DF signal is associated with local distribution of small angle scatterers in an image object, while the absorption signal and DPC signal are often used to characterize the relatively uniform structure of the image object. Besides the endogenous image contrast, exogenous contrast media are often used in x-ray imaging to locally enhance the image signal. This paper proposes a potential contrast medium for DF signal enhancement: microbubbles (MBs). MBs have already been developed for clinical use in ultrasound imaging, and recent experimental studies have shown that MBs may also enhance the DF signal, although it remained unclear how the physical characteristics of the MBs quantitatively impact the DF signal. In this paper, a systematic study was performed to investigate the quantitative relationships between the DF signal and the following properties of MBs: size, concentration, shell thickness, size uniformity, and whether gold nanoparticles were attached. The experimental results demonstrated that, an increased MB size (about 4 microns) may generate a stronger DF signal for our DPC imaging system; additionally, a moderately increased shell thickness and the use of gold nanoparticles on the shell surface also resulted in further enhancement of the DF signal. These findings may provide critical information needed for using MBs as the contrast agent of x-ray DF imaging.

X-ray grating-based phase-contrast Talbot-Lau interferometry is a promising imaging technology that has the potential to raise soft tissue contrast in comparison to conventional attenuation-based imaging. Additionally, it is sensitive to attenuation, refraction and scattering of the radiation and thus provides complementary and otherwise inaccessible information due to the dark-field image, which shows the sub-pixel size granularity of the measured object. Until recent progress the method has been mainly limited to photon energies below 40 keV. Scaling the method to photon energies that are sufficient to pass large and spacious objects represents a challenging task. This is caused by increasing demands regarding the fabrication process of the gratings and the broad spectra that come along with the use of polychromatic X-ray sources operated at high acceleration voltages. We designed a setup that is capable to reach high visibilities in the range from 50 to 120 kV. Therefore, spacious and dense parts of the human body with high attenuation can be measured, such as a human knee. The authors will show investigations on the resulting attenuation, differential phase-contrast and dark-field images. The images experimentally show that X-ray grating-based phase-contrast radiography is feasible with highly absorbing parts of the human body containing massive bones.

X-ray phase contrast imaging has been investigated during the last two decades for potential benefits in soft tissue imaging. Long imaging time, high radiation dose and general measurement complexity involving motion of x-ray optical components have prevented the clinical translation of these methods. In all existing popular phase contrast imaging methods, multiple measurements per projection angle involving motion of optical components are required to achieve quantitatively accurate estimation of absorption, phase and differential phase. Recently we proposed an algorithmic approach to use spectral detection data in a phase contrast imaging setup to obtain absorption, phase and differential phase in a single-step. Our generic approach has been shown via simulations in all three types of phase contrast imaging: propagation, coded aperture and grating interferometry. While other groups have used spectral detector in phase contrast imaging setups, our proposed method is unique in outlining an approach to use this spectral data to simplify phase contrast imaging. In this abstract we show the first experimental proof of our single-shot phase retrieval using a Medipix3 photon counting detector in an edge illumination aperture (also referred to as coded aperture) phase contrast set up as well as for a free space propagation setup. Our preliminary results validate our new transport equation for edge illumination PCI and our spectral phase retrieval algorithm for both PCI methods being investigated. Comparison with simulations also point to excellent performance of Medipix3 built-in charge sharing correction mechanism.

Flat-panel cone-beam CT (FP-CBCT) is a promising imaging modality, partly due to its potential for high spatial resolution reconstructions in relatively compact scanners. Despite this potential, FP-CBCT can face difficulty resolving important fine scale structures (e.g, trabecular details in dedicated extremities scanners and microcalcifications in dedicated CBCT mammography). Model-based methods offer one opportunity to improve high-resolution performance without any hardware changes. Previous work, based on a linearized forward model, demonstrated improved performance when both system blur and spatial correlations characteristics of FP-CBCT systems are modeled. Unfortunately, the linearized model relies on a staged processing approach that complicates tuning parameter selection and can limit the finest achievable spatial resolution. In this work, we present an alternative scheme that leverages a full nonlinear forward model with both system blur and spatially correlated noise. A likelihood-based objective function is derived from this forward model and we derive an iterative optimization algorithm for its solution. The proposed approach is evaluated in simulation studies using a digital extremities phantom and resolution-noise trade-offs are quantitatively evaluated. The correlated nonlinear model outperformed both the uncorrelated nonlinear model and the staged linearized technique with up to a 86% reduction in variance at matched spatial resolution. Additionally, the nonlinear models could achieve finer spatial resolution (correlated: 0.10 mm, uncorrelated: 0.11 mm) than the linear correlated model (0.15 mm), and traditional FDK (0.40 mm). This suggests the proposed nonlinear approach may be an important tool in improving performance for high-resolution clinical applications.

We present a new algorithm that allows for raw data-based automated cardiac and respiratory intrinsic gating in cone-beam CT scans. It can be summarized in three steps: First, a median filter is applied to an initially reconstructed volume. The forward projection of this volume contains less motion information and is subtracted from the original projections. This results in new raw data that contain only moving and not static anatomy like bones, that would otherwise impede the cardiac or respiratory signal acquisition. All further steps are applied to these modified raw data. Second, the raw data are cropped to a region of interest (ROI). The ROI in the raw data is determined by the forward projection of a binary volume of interest (VOI) that includes the diaphragm for respiratory gating and most of the edge of the heart for cardiac gating. Third, the mean gray value in this ROI is calculated for every projection and the respiratory/cardiac signal is acquired using a bandpass filter. Steps two and three are carried out simultaneously for 64 or 1440 overlapping VOI inside the body for the respiratory or cardiac signal respectively. The signals acquired from each ROI are compared and the most consistent one is chosen as the desired cardiac or respiratory motion signal. Consistency is assessed by the standard deviation of the time between two maxima. The robustness and efficiency of the method is evaluated using simulated and measured patient data by computing the standard deviation of the mean signal difference between the ground truth and the intrinsic signal.

Purpose: Prompt and reliable detection of intracranial hemorrhage (ICH) has substantial clinical impact in diagnosis and treatment of stroke and traumatic brain injury. This paper describes the design, development, and preliminary performance characterization of a dedicated cone-beam CT (CBCT) head scanner prototype for imaging of acute ICH. Methods: A task-based image quality model was used to analyze the detectability index as a function of system configuration, and hardware design was guided by the results of this model-based optimization. A robust artifact correction pipeline was developed using GPU-accelerated Monte Carlo (MC) scatter simulation, beam hardening corrections, detector veiling glare, and lag deconvolution. An iterative penalized weighted least-squares (PWLS) reconstruction framework with weights adjusted for artifact-corrected projections was developed. Various bowtie filters were investigated for potential dose and image quality benefits, with a MC-based tool providing estimates of spatial dose distribution. Results: The initial prototype will feature a source-detector distance of 1000 mm and source-axis distance of 550 mm, a 43x43 cm² flat panel detector, and a 15° rotating anode x-ray source with 15 kW power and 0.6 focal spot size. Artifact correction reduced image nonuniformity by ~250 HU, and PWLS reconstruction with modified weights improved the contrast to noise ratio by 20%. Inclusion of a bowtie filter can potentially reduce dose by 50% and improve CNR by 25%. Conclusions: A dedicated CBCT system capable of imaging millimeter-scale acute ICH was designed. Preliminary findings support feasibility of point-of-care applications in TBI and stroke imaging, with clinical studies beginning on a prototype.

In this work, a newly developed reconstruction algorithm, Synchronized MultiArtifact Reduction with Tomographic RECONstruction (SMART-RECON), was applied to C-arm cone beam CT perfusion (CBCTP) imaging. This algorithm contains a special rank regularizer, designed to reduce limited-view artifacts associated with super- short scan reconstructions. As a result, high temporal sampling and temporal resolution image reconstructions were achieved using an interventional C-arm x-ray system. The algorithm was evaluated in terms of the fidelity of the dynamic contrast update curves and the accuracy of perfusion parameters through numerical simulation studies. Results shows that, not only were the dynamic curves accurately recovered (relative root mean square error ∈ [3%, 5%] compared with [13%, 22%] for FBP), but also the noise in the final perfusion maps was dramatically reduced. Compared with filtered backprojection, SMART-RECON generated CBCTP maps with much improved capability in differentiating lesions with perfusion deficits from the surrounding healthy brain tissues.

Currently, clinical acquisition of IV 3D-DSA requires two separate scans: one mask scan without contrast medium and a filled scan with contrast injection. Having two separate scans adds radiation dose to the patient and increases the likelihood of suffering inadvertent patient motion induced mis-registration and the associated mis-registraion artifacts in IV 3D-DSA images. In this paper, a new technique, SMART-RECON is introduced to generate IV 3D-DSA images from a single Cone Beam CT (CBCT) acquisition to eliminate the mask scan. Potential benefits of eliminating mask scan would be: (1) both radiation dose and scan time can be reduced by a factor of 2; (2) intra-sweep motion can be eliminated; (3) inter-sweep motion can be mitigated. Numerical simulations were used to validate the algorithm in terms of contrast recoverability and the ability to mitigate limited view artifacts.

When an automatic exposure control is introduced in C-arm cone beam CT data acquisition, the spectral inconsistencies between acquired projection data are exacerbated. As a result, conventional water/bone correction schemes are not as effective as in conventional diagnostic x-ray CT acquisitions with a fixed tube potential. In this paper, a new method was proposed to reconstruct several images with different degrees of spectral consistency and thus different levels of beam hardening artifacts. The new method relies neither on prior knowledge of the x-ray beam spectrum nor on prior compositional information of the imaging object. Numerical simulations were used to validate the algorithm.

We introduce a novel strategy for fluence field modulation (FFM) in x-ray CT using multiple aperture devices (MADs). MAD filters permit FFM by blocking or transmitting the x-ray beam on a fine (0.1-1 mm) scale. The filters have a number of potential advantages over other beam modulation strategies including the potential for a highly compact design, modest actuation speed and acceleration requirements, and spectrally neutral filtration due to their essentially binary action. In this work, we present the underlying MAD filtration concept including a design process to achieve a specific class of FFM patterns. A set of MAD filters is fabricated using a tungsten laser sintering process and integrated into an x-ray CT test bench. A characterization of the MAD filters is conducted and compared to traditional attenuating bowtie filters and the ability to flatten the fluence profile for a 32 cm acrylic phantom is demonstrated. MAD-filtered tomographic data was acquired on the CT test bench and reconstructed without artifacts associated with the MAD filter. These initial studies suggest that MAD-based FFM is appropriate for integration in clinical CT system to create patient-specific fluence field profile and reduce radiation exposures.

Physicians rely on CT Perfusion (CTP) images and quantitative image data, including cerebral blood flow, cerebral blood volume, and bolus arrival delay, to diagnose and treat stroke patients. However, the quantification of these metrics may vary depending on the computational method used. Therefore, we have developed a dynamic and realistic digital brain phantom upon which CTP scans can be simulated based on a set of ground truth scenarios. Building upon the previously developed 4D extended cardiac-torso (XCAT) phantom containing a highly detailed brain model, this work consisted of expanding the intricate vasculature by semi-automatically segmenting existing MRA data and fitting nonuniform rational B-spline surfaces to the new vessels. Using time attenuation curves input by the user as reference, the contrast enhancement in the vessels changes dynamically. At each time point, the iodine concentration in the arteries and veins is calculated from the curves and the material composition of the blood changes to reflect the expected values. CatSim, a CT system simulator, generates simulated data sets of this dynamic digital phantom which can be further analyzed to validate CTP studies and post-processing methods. The development of this dynamic and realistic digital phantom provides a valuable resource with which current uncertainties and controversies surrounding the quantitative computations generated from CTP data can be examined and resolved.

Quantitative computed tomography (CT) applications based on statistical iterative reconstruction algorithms require accurate models of the CT acquisition process, with a key component being the x-ray fan beam intensity. We present a method to experimentally determine the extra-focal radiation profile incident on individual CT detectors. Using a tungsten cylinder as a knife edge, a super-sampled signal was created from sinogram data, which traced the “occlusion” of the x-ray source as seen by a detector. By differentiating this signal and correcting for finite detector size and motion blur, the effective source profile can be recovered. Extra-focal scatter was found to be on the order of 1-3 percent of the focal beam intensity, with lower relative magnitude at the isocenter and increasing towards the edge of the fan beam, with its profile becoming asymmetric at large angles. The implications for reconstruction algorithms and QCT applications will be discussed.

Cerebral CT perfusion (CTP) imaging is playing an important role in the diagnosis and treatment of acute ischemic strokes. Meanwhile, the reliability of CTP-based ischemic lesion detection has been challenged due to the noisy appearance and low signal-to-noise ratio of CTP maps. To reduce noise and improve image quality, a rigorous study on the noise transfer properties of CTP systems is highly desirable to provide the needed scientific guidance. This paper concerns how noise in the CTP source images propagates to the final CTP maps. Both theoretical deviations and subsequent validation experiments demonstrated that, the noise level of background frames plays a dominant role in the noise of the cerebral blood volume (CBV) maps. This is in direct contradiction with the general belief that noise of non-background image frames is of greater importance in CTP imaging. The study found that when radiation doses delivered to the background frames and to all non-background frames are equal, lowest noise variance is achieved in the final CBV maps. This novel equality condition provides a practical means to optimize radiation dose delivery in CTP data acquisition: radiation exposures should be modulated between background frames and non-background frames so that the above equality condition is satisïnAed. For several typical CTP acquisition protocols, numerical simulations and in vivo canine experiment demonstrated that noise of CBV can be effectively reduced using the proposed exposure modulation method.

The overall goal of this work is to develop a rapid, accurate and fully automated software tool to estimate patient-specific organ doses from computed tomography (CT) scans using a deterministic Boltzmann Transport Equation solver and automated CT segmentation algorithms. This work quantified the accuracy of organ dose estimates obtained by an automated segmentation algorithm. The investigated algorithm uses a combination of feature-based and atlas-based methods. A multiatlas approach was also investigated. We hypothesize that the auto-segmentation algorithm is sufficiently accurate to provide organ dose estimates since random errors at the organ boundaries will average out when computing the total organ dose. To test this hypothesis, twenty head-neck CT scans were expertly segmented into nine regions. A leave-one-out validation study was performed, where every case was automatically segmented with each of the remaining cases used as the expert atlas, resulting in nineteen automated segmentations for each of the twenty datasets. The segmented regions were applied to gold-standard Monte Carlo dose maps to estimate mean and peak organ doses. The results demonstrated that the fully automated segmentation algorithm estimated the mean organ dose to within 10% of the expert segmentation for regions other than the spinal canal, with median error for each organ region below 2%. In the spinal canal region, the median error was 7% across all data sets and atlases, with a maximum error of 20%. The error in peak organ dose was below 10% for all regions, with a median error below 4% for all organ regions. The multiple-case atlas reduced the variation in the dose estimates and additional improvements may be possible with more robust multi-atlas approaches. Overall, the results support potential feasibility of an automated segmentation algorithm to provide accurate organ dose estimates.

Energy-resolved photon-counting CT (PCCT) is promising for material decomposition with multi-contrast agents. However, corrections for non-idealities of PCCT detectors are required, which are still active research areas. In addition, PCCT is associated with very high cost due to lack of mass production. In this work, we proposed an alternative approach to performing multi-energy CT, which was achieved by acquiring triple or quadruple x-ray beam measurements on a dual-source CT scanner. This strategy was based on a “Twin Beam” design on a single-source scanner for dual-energy CT. Examples of beam filters and spectra for triple and quadruple x-ray beam were provided. Computer simulation studies were performed to evaluate the accuracy of material decomposition for multi-contrast mixtures using both tri-beam and quadruple-beam configurations. The proposed strategy can be readily implemented on a dual-source scanner, which may allow material decomposition of multi-contrast agents to be performed on clinical CT scanners with energy-integrating detector.

Respiration-correlated CT (4DCT) forms the basis of clinical 4D radiotherapy workflows for patients with thoracic and abdominal lesions. 4DCT image data, however, often suffers from motion artifacts due to unfulfilled assumptions during reconstruction and image/projection data sorting. In this work and focusing on low-pitch helical scanning protocols, two questionable assumptions are addressed: (1) the need for regular breathing patterns and (2) a constant correlation between the external breathing signal acquired for image/projection sorting and internal motion patterns. To counteract (1), a patient-specific upper breathing signal amplitude threshold is introduced to avoid artifacts due to unusual deep inspiration (helpful for both amplitude- and phase-based reconstruction). In addition, a projection data binning algorithm based on a statistical analysis of the patient's breathing signal is proposed to stabilize phase-based sorting. To further alleviate the need for (2), an image artifact metric is incorporated into and minimized during the reconstruction process. The optimized reconstruction is evaluated using 30 clinical 4DCT data sets and demonstrated to significantly reduce motion artifacts.

Detective Quantum Efficiency (DQE) is one of the most important image quality metrics for evaluating the spatial resolution performance of flat-panel x-ray detectors. In this work, we simulate the DQE of amorphous selenium (a-Se) xray detectors with a detailed Monte Carlo transport code (ARTEMIS) for modeling semiconductor-based direct x-ray detectors. The transport of electron-hole pairs is achieved with a spatiotemporal model that accounts for recombination and trapping of carriers and Coulombic effects of space charge and external applied electric field. A range of x-ray energies has been simulated from 10 to 100 keV. The DQE results can be used to study the spatial resolution characteristics of detectors at different energies.

Indirect conversion flat panel detectors (FPDs) based on amorphous silicon (a-Si) technology are widely used in digital X-ray imaging. In such FPDs a scintillator layer is used for converting X-rays into visible light photons. However, the lateral spread of these photons inside the scintillator layer reduces spatial resolution of the FPD. In this study, FPDs incorporating pixelated scintillators with a barrier rib structure were developed to limit lateral spread of light photons thereby improving spatial resolution. For the pixelated scintillator, a two-dimensional barrier rib structure was first manufactured on a substrate layer, coated with reflective materials, and filled to the rim with the scintillating material of gadolinium oxysulfide (GOS). Several scintillator samples were fabricated, with pitch size varying from 160 to 280 μm and rib height from 200 to 280 μm. The samples were directly coupled to an a-Si flat panel photodiode array with a pitch of 200 μm to convert optical photons to electronic signals. With the pixelated scintillator, the detector modulation transfer function was shown to improve significantly (by 94% at 2 cycle/mm) compared to a detector using an unstructured GOS layer. However, the prototype does show lower sensitivity due to the decrease in scintillator fill factor. The preliminary results demonstrated the feasibility of using the barrier-rib structure to improve the spatial resolution of FPDs. Such an improvement would greatly benefit nondestructive testing applications where the spatial resolution is the most important parameter. Further investigation will focus on improving the detector sensitivity and exploring its medical applications.

Digital Breast Tomosynthesis (DBT) requires excellent image quality in a dynamic mode at very low dose levels while Full Field Digital Mammography (FFDM) is a static imaging modality that requires high saturation dose levels. These opposing requirements can only be met by a dynamic detector with a high dynamic range. This paper will discuss a wafer-scale CMOS-based mammography detector with 49.5 μm pixels and a CsI scintillator. Excellent image quality is obtained for FFDM as well as DBT applications, comparing favorably with a-Se detectors that dominate the X-ray mammography market today. The typical dynamic range of a mammography detector is not high enough to accommodate both the low noise and the high saturation dose requirements for DBT and FFDM applications, respectively. An approach based on gain switching does not provide the signal-to-noise benefits in the low-dose DBT conditions. The solution to this is to add frame summing functionality to the detector. In one X-ray pulse several image frames will be acquired and summed. The requirements to implement this into a detector are low noise levels, high frame rates and low lag performance, all of which are unique characteristics of CMOS detectors. Results are presented to prove that excellent image quality is achieved, using a single detector for both DBT as well as FFDM dose conditions. This method of frame summing gave the opportunity to optimize the detector noise and saturation level for DBT applications, to achieve high DQE level at low dose, without compromising the FFDM performance.

Active matrix flat panel imagers (AMFPI) have become the dominant detector technology for digital radiography and fluoroscopy. For low dose imaging, electronic noise from the amorphous silicon thin film transistor (TFT) array degrades imaging performance. We have fabricated the first prototype solid-state AMFPI using a uniform layer of avalanche amorphous selenium (a-Se) photoconductor to amplify the signal to eliminate the effect of electronic noise. We have previously developed a large area solid-state avalanche a-Se sensor structure referred to as High Gain Avalanche Rushing Photoconductor (HARP) capable of achieving gains of 75. In this work we successfully deposited this HARP structure onto a 24 x 30 cm² TFT array with a pixel pitch of 85 μm. An electric field (E_Se) up to 105 Vμm^-1 was applied across the a-Se layer without breakdown. Using the HARP layer as a direct detector, an X-ray avalanche gain of 15 ± 3 was achieved at E_Se = 105 Vμm^-1. In indirect mode with a 150 μm thick structured CsI scintillator, an optical gain of 76 ± 5 was measured at E_Se = 105 Vμm^-1. Image quality at low dose increases with the avalanche gain until the electronic noise is overcome at a constant exposure level of 0.76 mR. We demonstrate the success of a solid-state HARP X-ray imager as well as the largest active area HARP sensor to date.

The ability to improve visualization of structural information in digital radiography without increasing radiation exposures requires improved image quality across all spatial frequencies, especially at high frequencies. The detective quantum efficiency (DQE) as a function of spatial frequency quantifies image quality given by an x-ray detector. We present a method of increasing DQE at high spatial frequencies by improving the modulation transfer function (MTF) and reducing noise aliasing. The Apodized Aperature Pixel (AAP) design uses a detector with micro-elements to synthesize desired pixels and provide higher DQE than conventional detector designs. A cascaded system analysis (CSA) that incorporates x-ray interactions is used for comparison of the theoretical MTF, noise power spectrum (NPS), and DQE. Signal and noise transfer through the converter material is shown to consist of correlated an uncorrelated terms. The AAP design was shown to improve the DQE of both material types that have predominantly correlated transfer (such as CsI) and predominantly uncorrelated transfer (such as Se). Improvement in the MTF by 50% and the DQE by 100% at the sampling cut-off frequency is obtained when uncorrelated transfer is prevalent through the converter material. Optimizing high-frequency DQE results in improved image contrast and visualization of small structures and fine-detail.

Sequential CT studies present an excellent opportunity to apply prior-image-based reconstruction (PIBR) methods that leverage high-fidelity prior imaging studies to improve image quality and/or reduce x-ray exposure in subsequent studies. One major obstacle in using PIBR is that the initial and subsequent studies are often performed on different scanners (e.g. diagnostic CT followed by CBCT for interventional guidance); this results in mismatch in attenuation values due to hardware and software differences. While improved artifact correction techniques can potentially mitigate such differences, the correction is often incomplete. Here, we present an alternate strategy where the PIBR itself is used to mitigate these differences. We define a new penalty for the previously introduced PIBR called Reconstruction of Difference (RoD). RoD differs from many other PIBRs in that it reconstructs only changes in the anatomy (vs. reconstructing the current anatomy). Direct regularization of the difference image in RoD provides an opportunity to selectively penalize spatial frequencies of the difference image (e.g. low frequency differences associated with attenuation offsets and shading artifacts) without interfering with the variations in unchanged background image. We leverage this flexibility and introduce a novel regularization strategy using a generalized Fourier penalty within the RoD framework and develop the modified reconstruction algorithm. We evaluate the performance of the new approach in both simulation studies and in physical CBCT test-bench data. We find that generalized Fourier penalty can be highly effective in reducing low-frequency x-ray artifacts through selective suppression of spatial frequencies in the reconstructed difference image.

Until today, several software-based approaches to increase the temporal resolution in cardiac computed tomography by estimating motion vector fields (MVFs) have been developed. Thereunder, the majority are motion compensation algorithms, which estimate the MVFs employing a three-dimensional registration routine working on reconstructions of multiple cardiac phases.^{2, 6, 7, 12} We present an algorithm that requires nothing more than the data needed for a short scan reconstruction for motion estimation and motion-compensated reconstruction, which both are based on the reconstruction of volumes from a limited angular range.^{2, 3, 7, 8} Those partial angle reconstructions are centered at different time points during the short scan and have a temporal resolution of about 10ms each. The MVFs are estimated by a constrained cost function optimization routine employing a motion artifact measuring cost function. During optimization, the MVFs are applied directly by warping the partial angle reconstructions, and the motion compensation is established by simply adding the shifted images. In order to enforce smooth vector fields and keep the number of parameters low, the motion is modeled by a low degree polynomial. Furthermore, to find a good estimation of the MVFs even in phases with rapid cardiac motion, the constrained optimization is re-initialized multiple times. The algorithm is validated with the help of a simulation study and applied to patient data, where motion- compensated reconstructions are performed in various cardiac phases. We show that the image quality can be improved, also in more rapid cardiac phases due to re-initialization of the optimization routine.

Lack of access to projection data from patient CT scans is a major limitation for development and validation of new reconstruction algorithms. To meet this critical need, we are building a library of CT patient projection data in an open and vendor-neutral format, DICOM-CT-PD, which is an extended DICOM format that contains sinogram data, acquisition geometry, patient information, and pathology identification. The library consists of scans of various types, including head scans, chest scans, abdomen scans, electrocardiogram (ECG)-gated scans, and dual-energy scans. For each scan, three types of data are provided, including DICOM-CT-PD projection data at various dose levels, reconstructed CT images, and a free-form text file. Several instructional documents are provided to help the users extract information from DICOM-CT-PD files, including a dictionary file for the DICOM-CT-PD format, a DICOM-CT-PD reader, and a user manual. Radiologist detection performance based on the reconstructed CT images is also provided. So far 328 head cases, 228 chest cases, and 228 abdomen cases have been collected for potential inclusion. The final library will include a selection of 50 head, chest, and abdomen scans each from at least two different manufacturers, and a few ECG-gated scans and dual-source, dual-energy scans. It will be freely available to academic researchers, and is expected to greatly facilitate the development and validation of CT reconstruction algorithms.

Iterative reconstruction methods have become very popular and show the potential to reduce dose. We present a limit to the maximum dose reduction possible with new reconstruction algorithms obtained by analyzing the information content of the raw data, assuming the reconstruction algorithm does not have a priori knowledge about the object or correlations between pixels. This limit applies to the task of estimating the density of a lesion embedded in a known background object, where the shape of the lesion is known but its density is not. Under these conditions, the density of the lesion can be estimated directly from the raw data in an optimal manner. This optimal estimate will meet or outperform the performance of any reconstruction method operating on the raw data, under the condition that the reconstruction method does not introduce a priori information. The raw data bound can be compared to the lesion density estimate from FBP in order to produce a limit on the dose reduction possible from new reconstruction algorithms. The possible dose reduction from iterative reconstruction varies with the object, but for a lesion embedded in the center of a water cylinder, it is less than 40%. Additionally, comparisons between iterative reconstruction and filtered backprojection are sometimes confounded by the effect of through-slice blurring in the iterative reconstruction. We analyzed the magnitude of the variance reduction brought about by through-slice blurring on scanners from two different vendors and found it to range between 11% and 48%.

When the scan field of view (SFOV) of a CT system is not large enough to enclose the entire cross-section of a patient, or the patient needs to be intentionally positioned partially outside the SFOV for certain clinical CT scans, truncation artifacts are often observed in the reconstructed CT images. Conventional wisdom to reduce truncation artifacts is to complete the truncated projection data via data extrapolation with different a priori assumptions. This paper presents a novel truncation artifact reduction method that directly works in the CT image domain. Specifically, a discriminative dictionary that includes a sub-dictionary of truncation artifacts and a sub-dictionary of non-artifact image information was used to separate a truncation artifact-contaminated image into two sub-images, one with reduced truncation artifacts, and the other one containing only the truncation artifacts. Both experimental phantom and retrospective human subject studies have been performed to characterize the performance of the proposed truncation artifact reduction method.

We present a novel reconstruction for motion correction of non-cardiac organs. With non-cooperative patients or in emergency case, breathing motion or motion of the skull may compromise image quality. Our algorithm is based on the optimization of either motion artefact metrics or data-driven metrics. This approach was successfully applied in cardiac CTA [1]. While motion correction of the coronary vessels requires a local motion model, global motion models are sufficient for organs like the lung or the skull. The parameter vector for the global affine motion is estimated iteratively, using the open source optimization library NLOPT. The image is updated using motion compensated reconstruction in each of the iterations. Evaluation of the metric value, e.g. the image entropy, provides information for the next iteration loop. After reaching the fixed point of the iteration, the final motion parameters are used for a motion-compensated full quality reconstruction. In head imaging the motion model is based on translation and rotation, in thoracic imaging the rotation is replaced by non-isotropic scaling in all three dimensions. We demonstrate the efficiency of the method in thoracic imaging by evaluating PET-CT data from free-breathing patients. In neuro imaging, data from stroke patients showing skull tremor were analyzed. It was shown that motion artefacts can be largely reduced and spatial resolution was restored. In head imaging, similar results can be obtained using motion artefact metrics or data-driven metrics. In case of image-based metrics, the entropy of the image proved to be superior. Breathing motion could also be significantly reduced using entropy metric. However, in this case data driven metrics cannot be applied because the line integrals associated to the ROI of the lung have to be computed using the local ROI mechanism [2] It was shown that the lung signal is corrupted by signals originating from the complement of the lung. Thus a meaningful optimization of a data-driven cost function is not possible.

Photon-counting CT (PCCT) may yield potential value for many clinical applications due to its relative immunity to electronic noise, increased geometric efficiency relative to current scintillating detectors, and the ability to resolve energy information about the detected photons. However, there are a large number of parameters that require optimization, particularly the energy thresholds configuration. Fast and accurate estimation of signal and noise in PCCT can benefit the optimization of acquisition parameters for specific diagnostic tasks. Based on the acquisition parameters and detector response of our research PCCT system, we derived mathematical models for both signal and noise. The signal model took the tube spectrum, beam filtration, object attenuation, water beam hardening, and detector response into account. The noise model considered the relationship between noise and radiation dose, as well as the propagation of noise as threshold data are subtracted to yield energy bin data. To determine the absolute noise value, a noise look-up table (LUT) was acquired using a limited number of calibration scans. The noise estimation algorithm then used the noise LUT to estimate noise for scans with a variety of combination of energy thresholds, dose levels, and object attenuation. Validation of the estimation algorithms was performed on our whole-body research PCCT system using semianthropomorphic water phantoms and solutions of calcium and iodine. The algorithms achieved accurate estimation of signal and noise for a variety of scanning parameter combinations. The proposed method can be used to optimize energy thresholds configuration for many clinical applications of PCCT.

The energy resolving capabilities of Photon Counting Detectors (PCD) in Computed Tomography (CT) facilitate energy-sensitive measurements. The provided image-information can be processed with Dual Energy and Multi Energy algorithms. A research PCD-CT firstly allows acquiring images with a close to clinical configuration of both the X-ray tube and the CT-detector. In this study, two algorithms (Material Decomposition and Virtual Non-Contrast-imaging (VNC)) are applied on a data set acquired from an anesthetized rabbit scanned using the PCD-CT system. Two contrast agents (CA) are applied: A gadolinium (Gd) based CA used to enhance contrasts for vascular imaging, and xenon (Xe) and air as a CA used to evaluate local ventilation of the animal's lung. Four different images are generated: a) A VNC image, suppressing any traces of the injected Gd imitating a native scan, b) a VNC image with a Gd-image as an overlay, where contrast enhancements in the vascular system are highlighted using colored labels, c) another VNC image with a Xe-image as an overlay, and d) a 3D rendered image of the animal's lung, filled with Xe, indicating local ventilation characteristics. All images are generated from two images based on energy bin information. It is shown that a modified version of a commercially available dual energy software framework is capable of providing images with diagnostic value obtained from the research PCD-CT system.

Shot noise processes are omnipresent in physics and many of their properties have been extensively studied in the past, including the particular problem of level crossing of shot noise. Energy-sensitive, photon-counting detectors using comparators to discriminate pulse-heights are currently heavily investigated for medical applications, e.g. for x-ray computed tomography and x-ray mammography. Surprisingly, no mention of the close relation between the two topics can be found in the literature on photon-counting detectors. In this paper, we point out the close analogy between level crossing of shot noise and the problem of determining count rates of photon- counting detectors subject to pulse pile-up. The latter is very relevant for obtaining precise forward models for photon-counting detectors operated under conditions of very high x-ray flux employed in clinical x-ray computed tomography. Although several attempts have been made to provide reasonably accurate, approximative models for the registered number of counts in x-ray detectors under conditions of high flux and arbitrary x-ray spectra, see, e.g., no exact, analytic solution is given in the literature for general continuous pulse shapes. In this paper we present such a solution for arbitrary response functions, x-ray spectra and continuous pulse shapes based on a result from the theory of level crossing. We briefly outline the theory of level crossing including the famous Rice theorem and translate from the language of level crossing to the language of photon-counting detection.

A high-resolution (HR) data collection mode has been introduced to a whole-body, research photon-counting-detector CT system installed in our laboratory. In this mode, 64 rows of 0.45 mm x 0.45 mm detector pixels were used, which corresponded to a pixel size of 0.25 mm x 0.25 mm at the iso-center. Spatial resolution of this HR mode was quantified by measuring the MTF from a scan of a 50 micron wire phantom. An anthropomorphic lung phantom, cadaveric swine lung, temporal bone and heart specimens were scanned using the HR mode, and image quality was subjectively assessed by two experienced radiologists. High spatial resolution of the HR mode was evidenced by the MTF measurement, with 15 lp/cm and 20 lp/cm at 10% and 2% modulation. Images from anthropomorphic phantom and cadaveric specimens showed clear delineation of small structures, such as lung vessels, lung nodules, temporal bone structures, and coronary arteries. Temporal bone images showed critical anatomy (i.e. stapes superstructure) that was clearly visible in the PCD system. These results demonstrated the potential application of this imaging mode in lung, temporal bone, and vascular imaging. Other clinical applications that require high spatial resolution, such as musculoskeletal imaging, may also benefit from this high resolution mode.

With many attractive attributes, photon counting detectors with many energy bins are being considered for clinical CT systems. In practice, a large amount of projection data acquired for multiple energy bins must be transferred in real time through slip rings and data storage subsystems, causing a bandwidth bottleneck problem. The higher resolution of these detectors and the need for faster acquisition additionally contribute to this issue. In this work, we introduce a new approach to lossless compression, specifically for projection data from photon counting detectors, by utilizing the dependencies in the multi-energy data. The proposed predictor estimates the value of a projection data sample as a weighted average of its neighboring samples and an approximation from other energy bins, and the prediction residuals are then encoded. Context modeling using three or four quantized local gradients is also employed to detect edge characteristics of the data. Using three simulated phantoms including a head phantom, compression of 2.3:1-2.4:1 was achieved. The proposed predictor using zero, three, and four gradient contexts was compared to JPEG-LS and the ideal predictor (noiseless projection data). Among our proposed predictors, three-gradient context is preferred with a compression ratio from Golomb coding 7% higher than JPEG-LS and only 3% lower than the ideal predictor. In encoder efficiency, the Golomb code with the proposed three-gradient contexts has higher compression than block floating point. We also propose a lossy compression scheme, which quantizes the prediction residuals with scalar uniform quantization using quantization boundaries that limit the ratio of quantization error variance to quantum noise variance. Applying our proposed predictor with three-gradient context, the lossy compression achieved a compression ratio of 3.3:1 but inserted a 2.1% standard deviation of error compared to that of quantum noise in reconstructed images. From the initial simulation results, the proposed algorithm shows good control over the bits needed to represent multienergy projection data.

There exists interest in designing a PET system with reduced detectors due to cost concerns, while not significantly compromising the PET utility. Recently developed optimization-based algorithms, which have demonstrated the potential clinical utility in image reconstruction from sparse CT data, may be used for enabling such design of innovative PET systems. In this work, we investigate a PET configuration with reduced number of detectors, and carry out preliminary studies from patient data collected by use of such sparse-PET configuration. We consider an optimization problem combining Kullback-Leibler (KL) data fidelity with an image TV constraint, and solve it by using a primal-dual optimization algorithm developed by Chambolle and Pock. Results show that advanced algorithms may enable the design of innovative PET configurations with reduced number of detectors, while yielding potential practical PET utilities.

The optic nerve (ON) plays a crucial role in human vision transporting all visual information from the retina to the brain for higher order processing. There are many diseases that affect the ON structure such as optic neuritis, anterior ischemic optic neuropathy and multiple sclerosis. Because the ON is the sole pathway for visual information from the retina to areas of higher level processing, measures of ON damage have been shown to correlate well with visual deficits. Increased intracranial pressure has been shown to correlate with the size of the cerebrospinal fluid (CSF) surrounding the ON. These measures are generally taken at an arbitrary point along the nerve and do not account for changes along the length of the ON. We propose a high contrast and high-resolution 3-D acquired isotropic imaging sequence optimized for ON imaging. We have acquired scan-rescan data using the optimized sequence and a current standard of care protocol for 10 subjects. We show that this sequence has superior contrast-to-noise ratio to the current standard of care while achieving a factor of 11 higher resolution. We apply a previously published automatic pipeline to segment the ON and CSF sheath and measure the size of each individually. We show that these measures of ON size have lower short- term reproducibility than the population variance and the variability along the length of the nerve. We find that the proposed imaging protocol is (1) useful in detecting population differences and local changes and (2) a promising tool for investigating biomarkers related to structural changes of the ON.

In time-of-flight (TOF) positron emission tomography (PET), precise calibration of the timing offset of each crystal of a PET scanner is essential. Conventionally this calibration requires a specially designed tool just for this purpose. In this study a method that uses a planar source to measure the crystal timing offsets (CTO) is developed. The method uses list mode acquisitions of a planar source placed at multiple orientations inside the PET scanner field-of-view (FOV). The placement of the planar source in each acquisition is automatically figured out from the measured data, so that a fixture for exactly placing the source is not required. The expected coincidence time difference for each detected list mode event can be found from the planar source placement and the detector geometry. A deviation of the measured time difference from the expected one is due to CTO of the two crystals. The least squared solution of the CTO is found iteratively using the list mode events. The effectiveness of the crystal timing calibration method is evidenced using phantom images generated by placing back each list mode event into the image space with the timing offset applied to each event. The zigzagged outlines of the phantoms in the images become smooth after the crystal timing calibration is applied. In conclusion, a crystal timing calibration method is developed. The method uses multiple list mode acquisitions of a planar source to find the least squared solution of crystal timing offsets.

Unaccounted scatter impact from unknown outside-axial-field-of-view (outside-AFOV) activity in PET is an important degrading factor for image quality and quantitation. Resource consuming and unpopular way to account for the outside- AFOV activity is to perform an additional PET/CT scan of adjacent regions. In this work we investigate a solution to the outside-AFOV scatter problem without performing a PET/CT scan of the adjacent regions. The main motivation for the proposed method is that the measured random corrected prompt (RCP) sinogram in the background region surrounding the measured object contains only scattered events, originating from both inside- and outside-AFOV activity. In this method, the scatter correction simulation searches through many randomly-chosen outside-AFOV activity estimates along with known inside-AFOV activity, generating a plethora of scatter distribution sinograms. This digital experimentation iterates until a decent match is found between a simulated scatter sinogram (that include supposed outside-AFOV activity) and the measured RCP sinogram in the background region. The combined scatter impact from inside- and outside-AFOV activity can then be used for scatter correction during final image reconstruction phase. Preliminary results using measured phantom data indicate successful phantom length estimate with the method, and, therefore, accurate outside-AFOV scatter estimate.

Photon counting detectors in computed tomography facilitate measurements of spectral distributions of detected X-ray quanta in discrete energy bins. Along with the dependency on wavelength and atomic number of the mass attenuation coefficient, this information allows for reconstruction of CT images of different material bases. Decomposition of two materials is considered standard in today’s dual-energy techniques. With photon-counting detectors the decomposition of more than two materials becomes achievable. Efficient detection of CT-typical X-ray spectra is a hard requirement in a clinical environment. This is fulfilled by only a few sensor materials such as CdTe or CdZnTe. In contrast to energy integrating CT-detectors, the pixel dimensions must be reduced to avoid pulse pile-up problems at clinically relevant count rates. However, reducing pixel sizes leads to increased K-escape and charge sharing effects. As a consequence, the correlation between incident and detected X-ray energy is reduced. This degradation is quantified by the detector response function. The goal of this study is to improve the achievable material decomposition by adapting the incident X-ray spectrum with respect to the properties (i.e. the detector response function) of a photon counting detector. A significant improvement of a material decomposition equivalent metric is achievable when using specific materials as X-ray pre-filtration (K-edge filtering) while maintaining the applied patient dose and image quality.

Recent advances in Photon Counting CT (PCCT) have facilitated the simultaneous acquisition of multiple image volumes with differing energy thresholds. This presents the user with several choices for energy threshold combinations. As compared to standard clinical Dual kVp CT , where the user typically has only three choices of kVp pairings (e.g., 80/150Sn, 90/150Sn, 100/150Sn), a “quad” PCCT system with 14 threshold settings has Choose(14,4)= 1001 possible threshold combinations (assuming no restrictions). In this paper we describe a computationally tractable means to order, from best (most accurate) to worst (least accurate), threshold combinations for the task of discriminating pure materials of assumed approximate concentrations using the Bhattacharyya Coefficient. We observe that this ordering is not necessarily identical to the ordering for the task of decomposing material mixtures into their components. We demonstrate our approach on phantom data.

Spectral imaging systems need to be able to produce "conventional" images, and it's been shown that systems with energy discriminating detectors can achieve higher CNR than conventional systems by optimal weighting. Combining measured data in energy bins (EBs) and also combining basis material images have previously been proposed, but there are no studies systematically comparing the two methods. In this paper, we analytically evaluate the two methods for systems with ideal photon counting detectors using CNR and beam hardening (BH) artifact as metrics. For a 120-kVp polychromatic simulations of a water phantom with low contrast inserts, the difference of the optimal CNR between the two methods for the studied phantom is within 2%. For a polychromatic spectrum, beam-hardening artifacts are noticeable in EB weighted images (BH artifact of 3.8% for 8 EB and 6.9% for 2 EB), while weighted basis material images are free of such artifacts.

An x-ray photon interacts with photon counting detectors (PCDs) and generates an electron charge cloud or multiple clouds. The clouds (thus, the photon energy) may be split between two adjacent PCD pixels when the interaction occurs near pixel boundaries, producing a count at both of the two pixels. This is called double-counting with charge sharing. The output of individual PCD pixel is Poisson distributed integer counts; however, the outputs of adjacent pixels are correlated due to double-counting. Major problems are the lack of detector noise model for the spatio-energetic crosstalk and the lack of an efficient simulation tool. Monte Carlo simulation can accurately simulate these phenomena and produce noisy data; however, it is not computationally efficient. In this study, we developed a new detector model and implemented into an efficient software simulator which uses a Poisson random number generator to produce correlated noisy integer counts. The detector model takes the following effects into account effects: (1) detection efficiency and incomplete charge collection; (2) photoelectric effect with total absorption; (3) photoelectric effect with fluorescence x-ray emission and re-absorption; (4) photoelectric effect with fluorescence x-ray emission which leaves PCD completely; and (5) electric noise. The model produced total detector spectrum similar to previous MC simulation data. The model can be used to predict spectrum and correlation with various different settings. The simulated noisy data demonstrated the expected performance: (a) data were integers; (b) the mean and covariance matrix was close to the target values; (c) noisy data generation was very efficient

Using an energy-resolving photon counting detector, the amount of k-edge material in the x-ray path can be estimated using a process known as material decomposition. However, non-ideal effects within the detector make it difficult to accurately perform this decomposition. This work evaluated the k-edge material decomposition accuracy of two empirical estimators. A neural network estimator and a linearized maximum likelihood estimator with error look-up tables (A-table method) were evaluated through simulations and experiments. Each estimator was trained on system-specific calibration data rather than specific modeling of non-ideal detector effects or the x-ray source spectrum. Projections through a step-wedge calibration phantom consisting of different path lengths through PMMA, aluminum, and a k-edge material was used to train the estimators. The estimators were tested by decomposing data acquired through different path lengths of the basis materials. The estimators had similar performance in the chest phantom simulations with gadolinium. They estimated four of the five densities of gadolinium with less than 2mg/mL bias. The neural networks estimates demonstrated lower bias but higher variance than the A-table estimates in the iodine contrast agent simulations. The neural networks had an experimental variance lower than the CRLB indicating it is a biased estimator. In the experimental study, the k-edge material contribution was estimated with less than 14% bias for the neural network estimator and less than 41% bias for the A-table method.

This study was performed to test the feasibility of multi-gamma-source CT imaging system. Gamma-source CT employs radioisotopes that emit monochromatic energy gamma-rays. The advantages of gamma-source CT include its immunity to beam hardening artifacts, its capacity of quantitative CT imaging, and its higher performance in low contrast imaging compared to the conventional x-ray CT. Radioisotope should be shielded by use of a pin-hole collimator so as to make a fine focal spot. Due to its low gamma-ray flux in general, the reconstructed image from a single gamma-source CT would suffer from high noise in data. To address this problem, we proposed a multi-gamma source CT imaging system and developed an iterative image reconstruction algorithm accordingly in this work. Conventional imaging model assumes a single linear imaging system typically represented by Mf = g. In a multi-gamma-source CT system however, the inversion problem is not any more based on a single linear system since one cannot separate a detector pixel value into multiple ones that are corresponding to each rays from the sources. Instead, the imaging model can be constructed by a set of linear system models each of which assumes an estimated measurement g. Based on this model, the proposed algorithm has a weighting step which distributes each projection data into multiple estimated measurements. We used two gamma sources at various positions and with varying intensities in this numerical study to demonstrate its feasibility. Therefore, the measured projection data(g) is separated into each estimated projection data(g₁, g₂) in this study. The proposed imaging protocol is believed to contribute to both medical and industrial applications.

X-ray fluorescence tomography (XFCT) has potential for high-resolution 3D molecular x-ray bio-imaging. In this technique the fluorescence signal from targeted nanoparticles (NPs) is measured, providing information about the spatial distribution and concentration of the NPs inside the object. However, present laboratory XFCT systems typically have limited spatial resolution (>1 mm) and suffer from long scan times and high radiation dose even at high NP concentrations, mainly due to low efficiency and poor signal-to-noise ratio. We have developed a laboratory XFCT system with high spatial resolution (sub-100 μm), low NP concentration and vastly decreased scan times and dose, opening up the possibilities for in-vivo small-animal imaging research. The system consists of a high-brightness liquid-metal-jet microfocus x-ray source, x-ray focusing optics and an energy-resolving photon-counting detector. By using the source’s characteristic 24 keV line-emission together with carefully matched molybdenum nanoparticles the Compton background is greatly reduced, increasing the SNR. Each measurement provides information about the spatial distribution and concentration of the Mo nanoparticles. A filtered back-projection method is used to produce the final XFCT image.

A small animal micro-CT system was built using an EMCCD detectors having complex pre-digitization amplification technology, high-resolution, high-sensitivity and low-noise. Noise in CBCT reconstructed images when using predigitization amplification behaves differently than commonly used detectors and warrants a detailed investigation. In this study, noise power and contrast sensitivity were estimated for the newly built system. Noise analysis was performed by scanning a water phantom. Tube voltage was lowered to minimum delivered by the tube (20 kVp and 0.5 mA) and detector gain was varied. Contrast sensitivity was analyzed by using a phantom containing different iodine contrast solutions (20% to 70%) filled in six different tubes. First, we scanned the phantom using various x-ray exposures at 40 kVp while changing the gain to maintain the background air value of the projection images constant. Next, the exposure was varied while the detector gain was maintained constant. Radial NPS plots show that noise power level increases as gain increases. Contrast sensitivity was analyzed by calculating ratio of signal-to-noise ratios (SNR) for increased gain with those of low constant gain at each exposure. The SNR value at low constant gain was always lower than SNR of high detector gain at all x-ray settings and iodine contrast. The largest increase of SNR approached 1.3 for low contrast feature for an iodine concentration of 20%. Despite an increase in noise level as gain increases, the SNR improvement shows that signal level also increases because of the unique on-chip gain of the detector.

Proton radiography shows great promise as a tool to guide proton beam therapy (PBT) in real time. Here, we demonstrate two ways in which the technology may progress towards that goal. Firstly, with a proton beam that is 800 MeV in energy, target tissue receives a dose of radiation with very tight lateral constraint. This could present a benefit over the traditional treatment energies of ~200 MeV, where up to 1 cm of lateral tissue receives scattered radiation at the target. At 800 MeV, the beam travels completely through the object with minimal deflection, thus constraining lateral dose to a smaller area. The second novelty of this system is the utilization of magnetic quadrupole refocusing lenses that mitigate the blur caused by multiple Coulomb scattering within an object, enabling high resolution imaging of thick objects, such as the human body. This system is demonstrated on ex vivo salamander and zebrafish specimens, as well as on a realistic hand phantom. The resulting images provide contrast sufficient to visualize thin tissue, as well as fine detail within the target volumes, and the ability to measure small changes in density. Such a system, combined with PBT, would enable the delivery of a highly specific dose of radiation that is monitored and guided in real time.

Scanning-beam digital x-ray (SBDX) is an inverse geometry x-ray fluoroscopy system capable of tomosynthesis-based 3D catheter tracking. This work proposes a method of dose-reduced 3D tracking using dynamic electronic collimation (DEC) of the SBDX scanning x-ray tube. Positions in the 2D focal spot array are selectively activated to create a regionof- interest (ROI) x-ray field around the tracked catheter. The ROI position is updated for each frame based on a motion vector calculated from the two most recent 3D tracking results. The technique was evaluated with SBDX data acquired as a catheter tip inside a chest phantom was pulled along a 3D trajectory. DEC scans were retrospectively generated from the detector images stored for each focal spot position. DEC imaging of a catheter tip in a volume measuring 11.4 cm across at isocenter required 340 active focal spots per frame, versus 4473 spots in full-FOV mode. The dose-area-product (DAP) and peak skin dose (PSD) for DEC versus full field-of-view (FOV) scanning were calculated using an SBDX Monte Carlo simulation code. DAP was reduced to 7.4% to 8.4% of the full-FOV value, consistent with the relative number of active focal spots (7.6%). For image sequences with a moving catheter, PSD was 33.6% to 34.8% of the full-FOV value. The root-mean-squared-deviation between DEC-based 3D tracking coordinates and full-FOV 3D tracking coordinates was less than 0.1 mm. The 3D distance between the tracked tip and the sheath centerline averaged 0.75 mm. Dynamic electronic collimation can reduce dose with minimal change in tracking performance.

Small-angle x-ray scattering (SAXS) imaging is an emerging medical tool that can be used for in vivo detailed tissue characterization and has the potential to provide added contrast to conventional x-ray projection and CT imaging. We used a publicly available MC-GPU code to simulate x-ray trajectories in a SAXS-CT geometry for a target material embedded in a water background material with varying sample sizes (1, 3, 5, and 10 mm). Our target materials were water solution of gold nanoparticle (GNP) spheres with a radius of 6 nm and a water solution with dissolved serum albumin (BSA) proteins due to their well-characterized scatter profiles at small angles and highly scattering properties. The background material was water. Our objective is to study how the reconstructed scatter profile degrades at larger target imaging depths and increasing sample sizes. We have found that scatter profiles of the GNP in water can still be reconstructed at depths up to 5 mm embedded at the center of a 10 mm sample. Scatter profiles of BSA in water were also reconstructed at depths up to 5 mm in a 10 mm sample but with noticeable signal degradation as compared to the GNP sample. This work presents a method to study the sample size limits for future SAXS-CT imaging systems.

It is known that conventional x-ray imaging provides a maximum contrast between cancerous and healthy fibroglandular breast tissues of 3% based on their linear x-ray attenuation coefficients at 17.5 keV, whereas coherent scatter signal provides a maximum contrast of 19% based on their differential coherent scatter cross sections. Therefore in order to exploit this potential contrast, we seek to evaluate the performance of a coded- aperture coherent scatter imaging system for breast cancer detection and investigate its accuracy using Monte Carlo simulations. In the simulations we modeled our experimental system, which consists of a raster-scanned pencil beam of x-rays, a bismuth-tin coded aperture mask comprised of a repeating slit pattern with 2-mm periodicity, and a linear-array of 128 detector pixels with 6.5-keV energy resolution. The breast tissue that was scanned comprised a 3-cm sample taken from a patient-based XCAT breast phantom containing a tomosynthesis- based realistic simulated lesion. The differential coherent scatter cross section was reconstructed at each pixel in the image using an iterative reconstruction algorithm. Each pixel in the reconstructed image was then classified as being either air or the type of breast tissue with which its normalized reconstructed differential coherent scatter cross section had the highest correlation coefficient. Comparison of the final tissue classification results with the ground truth image showed that the coded aperture imaging technique has a cancerous pixel detection sensitivity (correct identification of cancerous pixels), specificity (correctly ruling out healthy pixels as not being cancer) and accuracy of 92.4%, 91.9% and 92.0%, respectively. Our Monte Carlo evaluation of our experimental coded aperture coherent scatter imaging system shows that it is able to exploit the greater contrast available from coherently scattered x-rays to increase the accuracy of detecting cancerous regions within the breast.

Coded aperture X-ray diffraction (coherent scatter spectral) imaging provides fast and dose-efficient measurements of the molecular structure of an object. The information provided is spatially-dependent and material-specific, and can be utilized in medical applications requiring material discrimination, such as tumor imaging. However, current coded aperture coherent scatter spectral imaging system assume a uniformly or weakly attenuating object, and are plagued by image degradation due to non-uniform self-attenuation. We propose accounting for such non-uniformities in the self-attenuation by utilizing an X-ray computed tomography (CT) image (reconstructed attenuation map). In particular, we present an iterative algorithm for coherent scatter spectral image reconstruction, which incorporates the attenuation map, at different stages, resulting in more accurate coherent scatter spectral images in comparison to their uncorrected counterpart. The algorithm is based on a spectrally grouped edge-preserving regularizer, where the neighborhood edge weights are determined by spatial distances and attenuation values.

Purpose: This work applies task-driven optimization to design CT tube current modulation and directional regularization in penalized-likelihood (PL) reconstruction. The relative performance of modulation schemes commonly adopted for filtered-backprojection (FBP) reconstruction were also evaluated for PL in comparison. Methods: We adopt a task-driven imaging framework that utilizes a patient-specific anatomical model and information of the imaging task to optimize imaging performance in terms of detectability index (d’). This framework leverages a theoretical model based on implicit function theorem and Fourier approximations to predict local spatial resolution and noise characteristics of PL reconstruction as a function of the imaging parameters to be optimized. Tube current modulation was parameterized as a linear combination of Gaussian basis functions, and regularization was based on the design of (directional) pairwise penalty weights for the 8 in-plane neighboring voxels. Detectability was optimized using a covariance matrix adaptation evolutionary strategy algorithm. Task-driven designs were compared to conventional tube current modulation strategies for a Gaussian detection task in an abdomen phantom. Results: The task-driven design yielded the best performance, improving d’ by ~20% over an unmodulated acquisition. Contrary to FBP, PL reconstruction using automatic exposure control and modulation based on minimum variance (in FBP) performed worse than the unmodulated case, decreasing d’ by 16% and 9%, respectively. Conclusions: This work shows that conventional tube current modulation schemes suitable for FBP can be suboptimal for PL reconstruction. Thus, the proposed task-driven optimization provides additional opportunities for improved imaging performance and dose reduction beyond that achievable with conventional acquisition and reconstruction.

Reader studies are important in assessing breast imaging systems. The purpose of this work was to assess task-based performance of full field digital mammography (FFDM), digital breast tomosynthesis (DBT), and synthetic mammography (SM) using different phantom types, and to determine an accurate observer model for human readers.

Images were acquired on a Hologic Selenia Dimensions system with a uniform and anthropomorphic phantom. A contrast detail insert of small, low-contrast disks was created using an inkjet printer with iodine-doped ink and inserted in the phantoms. The disks varied in diameter from 210 to 630 μm, and in contrast from 1.1% contrast to 2.2% in regular increments. Human and model observers performed a 4-alternative forced choice experiment. The models were a non-prewhitening matched filter with eye model (NPWE) and a channelized Hotelling observer with either Gabor channels (Gabor-CHO) or Laguerre-Gauss channels (LG-CHO).

With the given phantoms, reader scores were higher in FFDM and DBT than SM. The structure in the phantom background had a bigger impact on outcome for DBT than for FFDM or SM. All three model observers showed good correlation with humans in the uniform background, with ρ between 0.89 and 0.93. However, in the structured background, only the CHOs had high correlation, with ρ=0.92 for Gabor-CHO, 0.90 for LG-CHO, and 0.77 for NPWE.

Because results of any analysis can depend on the phantom structure, conclusions of modality performance may need to be taken in the context of an appropriate model observer and a realistic phantom.

The use of Fourier domain model observer is challenged by iterative reconstruction (IR), because IR algorithms are nonlinear and IR images have noise texture different from that of FBP. A modified Fourier domain model observer, which incorporates nonlinear noise and resolution properties, has been proposed for IR and needs to be validated with human detection performance. On the other hand, the spatial domain model observer is theoretically applicable to IR, but more computationally intensive than the Fourier domain method. The purpose of this study is to compare the modified Fourier domain model observer to the spatial domain model observer with both FBP and IR images, using human detection performance as the gold standard. A phantom with inserts of various low contrast levels and sizes was repeatedly scanned 100 times on a third-generation, dual-source CT scanner at 5 dose levels and reconstructed using FBP and IR algorithms. The human detection performance of the inserts was measured via a 2-alternative-forced-choice (2AFC) test. In addition, two model observer performances were calculated, including a Fourier domain non-prewhitening model observer and a spatial domain channelized Hotelling observer. The performance of these two mode observers was compared in terms of how well they correlated with human observer performance. Our results demonstrated that the spatial domain model observer correlated well with human observers across various dose levels, object contrast levels, and object sizes. The Fourier domain observer correlated well with human observers using FBP images, but overestimated the detection performance using IR images.

Although digital mammography has reduced breast cancer mortality by approximately 30%, sensitivity and specificity are still far from perfect. In particular, the performance of mammography is especially limited for women with dense breast tissue. Two out of every three biopsies performed in the U.S. are unnecessary, thereby resulting in increased patient anxiety, pain, and possible complications. One promising tomographic breast imaging method that has recently been approved by the FDA is dedicated breast computed tomography (BCT). However, visualizing lesions with BCT can still be challenging for women with dense breast tissue due to the minimal contrast for lesions surrounded by fibroglandular tissue. In recent years there has been renewed interest in improving lesion conspicuity in x-ray breast imaging by administration of an iodinated contrast agent. Due to the fully 3-D imaging nature of BCT, as well as sub-optimal contrast enhancement while the breast is under compression with mammography and breast tomosynthesis, dedicated BCT of the uncompressed breast is likely to offer the best solution for injected contrast-enhanced x-ray breast imaging. It is well known that use of statistically-based iterative reconstruction in CT results in improved image quality at lower radiation dose. Here we investigate possible improvements in image reconstruction for BCT, by optimizing free regularization parameter in method of maximum likelihood and comparing its performance with clinical cone-beam filtered backprojection (FBP) algorithm.

In x-ray computed tomography (CT), task-based image quality studies are typically performed using uniform background phantoms with low-contrast signals. Such studies may have limited clinical relevancy for modern non-linear CT systems due to possible influence of background texture on image quality. The purpose of this study was to design and implement anatomically informed textured phantoms for task-based assessment of low-contrast detection. Liver volumes were segmented from 23 abdominal CT cases. The volumes were characterized in terms of texture features from gray-level co-occurrence and run-length matrices. Using a 3D clustered lumpy background (CLB) model, a fitting technique based on a genetic optimization algorithm was used to find the CLB parameters that were most reflective of the liver textures, accounting for CT system factors of spatial blurring and noise. With the modeled background texture as a guide, a cylinder phantom (165 mm in diameter and 30 mm height) was designed, containing 20 low-contrast spherical signals (6 mm in diameter at targeted contrast levels of ~3.2, 5.2, 7.2, 10, and 14 HU, 4 repeats per signal). The phantom was voxelized and input into a commercial multi-material 3D printer (Object Connex 350), with custom software for voxel-based printing. Using principles of digital half-toning and dithering, the 3D printer was programmed to distribute two base materials (VeroWhite and TangoPlus, nominal voxel size of 42x84x30 microns) to achieve the targeted spatial distribution of x-ray attenuation properties. The phantom was used for task-based image quality assessment of a clinically available iterative reconstruction algorithm (Sinogram Affirmed Iterative Reconstruction, SAFIRE) using a channelized Hotelling observer paradigm. Images of the textured phantom and a corresponding uniform phantom were acquired at six dose levels and observer model performance was estimated for each condition (5 contrasts x 6 doses x 2 reconstructions x 2 backgrounds = 120 total conditions). Based on the observer model results, the dose reduction potential of SAFIRE was computed and compared between the uniform and textured phantom. The dose reduction potential of SAFIRE was found to be 23% based on the uniform phantom and 17% based on the textured phantom. This discrepancy demonstrates the need to consider background texture when assessing non-linear reconstruction algorithms.

In previous work we developed a method for predicting the minimum detectable change (MDC) in nodule volume based on volumetric CT measurements. MDC was defined as the minimum increase/decrease in a nodule volume distinguishable from the baseline measurement at a specified level of detection performance, assessed using the area under the ROC curve (AUC). In this work we derived volume estimates of a set of synthetic nodules and calculated the detection performance for distinguishing them from baseline measurements. Eight spherical objects of 100HU radio density ranging in diameter from 5.0mm to 5.75mm and 8.0mm to 8.75mm with 0.25mm increments were placed in an anthropomorphic phantom with either no background (high-contrast task) or gelatin background (low-contrast task). The baseline was defined as 5.0mm for the first set of nodules and 8.0mm for the second set. The phantom was scanned using varying exposures, and reconstructed with slice thickness of 0.75, 1.5, and 3.0mm and two reconstruction kernels (standard and smooth). Volume measurements were derived using a previously developed matched- filter approach. Results showed that nodule size, slice thickness, and nodule-to-background contrast affected detectable change in nodule volume when using our volume estimator and the acquisition settings from our study. We also compared our experimental results to the values estimated by our previously-developed MDC prediction method. We found that experimental data for the 8mm baseline nodules matched very well with our predicted values of MDC. These results support considering the use of this metric when standardizing imaging protocols for lung nodule size change assessment.

To facilitate rigorous virtual clinical trials using model observers for breast imaging optimization and evaluation, we demonstrated a method of defining statistical models, based on 177 sets of breast CT patient data, in order to generate tens of thousands of unique digital breast phantoms. In order to separate anatomical texture from variation in breast shape, each training set of breast phantoms were deformed to a consistent atlas compressed geometry. Principal component analysis (PCA) was then performed on the shape-matched breast CT volumes to capture the variation of patient breast textures. PCA decomposes the training set of N breast CT volumes into an N-1-dimensional space of eigenvectors, which we call eigenbreasts. By summing weighted combinations of eigenbreasts, a large ensemble of different breast phantoms can be newly created. Different training sets can be used in eigenbreast analysis for designing basis models to target sub-populations defined by breast characteristics, such as size or density. In this work, we plan to generate ensembles of 30,000 new phantoms based on glandularity for an upcoming virtual trial of lesion detectability in digital breast tomosynthesis. Our method extends our series of digital and physical breast phantoms based on human subject anatomy, providing the capability to generate new, unique ensembles consisting of tens of thousands or more virtual subjects. This work represents an important step towards conducting future virtual trials for tasks-based assessment of breast imaging, where it is vital to have a large ensemble of realistic phantoms for statistical power as well as clinical relevance.

A new 3D breast computed tomography (CT) system is under development enabling imaging of microcalcifications in a fully uncompressed breast including posterior chest wall tissue. The system setup uses a steered electron beam impinging on small tungsten targets surrounding the breast to emit X-rays. A realization of the corresponding detector concept is presented in this work and it is modeled through Monte Carlo simulations in order to quantify first characteristics of transmission and secondary photons. The modeled system comprises a vertical alignment of linear detectors hold by a case that also hosts the breast. Detectors are separated by gaps to allow the passage of X-rays towards the breast volume. The detectors located directly on the opposite side of the gaps detect incident X-rays. Mechanically moving parts in an imaging system increase the duration of image acquisition and thus can cause motion artifacts. So, a major advantage of the presented system design is the combination of the fixed detectors and the fast steering electron beam which enable a greatly reduced scan time. Thereby potential motion artifacts are reduced so that the visualization of small structures such as microcalcifications is improved. The result of the simulation of a single projection shows high attenuation by parts of the detector electronics causing low count levels at the opposing detectors which would require a flat field correction, but it also shows a secondary to transmission ratio of all counted X-rays of less than 1 percent. Additionally, a single slice with details of various sizes was reconstructed using filtered backprojection. The smallest detail which was still visible in the reconstructed image has a size of 0.2mm.

Virtual clinical trials are a promising new approach increasingly used for the evaluation and comparison of breast imaging modalities. A key component in such an assessment paradigm is the use of simulated pathology, in particular, simulation of lesions. Breast mass lesions can be generally classified into two categories based on their appearance; nonspiculated masses and spiculated masses. In our previous work, we have successfully simulated non-spiculated masses using a fractal growth process known as diffusion limited aggregation. In this new work, we have extended the DLA model to simulate spiculated lesions by using features extracted from patient DBT images containing spiculated lesions. The features extracted included spicule length, width, curvature and distribution. This information was used to simulate realistic looking spicules which were attached to the surface of a DLA mass to produce a spiculated mass. A batch of simulated spiculated masses was inserted into normal patient images and presented to an experienced radiologist for review. The study yielded promising results with the radiologist rating 60% of simulated lesions in 2D and 50% of simulated lesions in DBT as realistic.

Energy resolved photon-counting mammography is a new technology, which counts the number of photons that passes through an object, and presents it as a pixel value in an image of the object. Silicon semiconductor detectors are currently used in commercial mammography. However, the disadvantage of silicon is the low absorption efficiency for high X-ray energies. A cadmium telluride (CdTe) series detector has a high absorption efficiency over a wide energy range. In this study, we proposed a method to estimate the composition of the mammary gland using a CdTe series detector as a photon-counting detector. The fact that the detection rate of breast cancer in mammography is affected by mammary gland composition is now widely accepted. Assessment of composition of the mammary gland has important implications. An important advantage of our proposed technique is its ability to discriminate photons using three energy bins. We designed the CdTe series detector system using the MATLAB simulation software. The phantom contains nine regions with the ratio of glandular tissue and adipose varying in increments of 10%. The attenuation coefficient for each bin’s energy was calculated from the number of input and output photons possessed by each. The evaluation results obtained by plotting the attenuation coefficient μ in a three-dimensional (3D) scatter plot show that the plots had a regular composition order congruent with that of the mammary gland. Consequently, we believe that our proposed method can be used to estimate the composition of the mammary gland.

We propose a new mammography system using a cadmium telluride (CdTe) series photon-counting detector, having high absorption efficiency over a wide energy range. In a previous study, we showed that the use of high X-ray energy in digital mammography is useful from the viewpoint of exposure dose and image quality. In addition, the CdTe series detector can acquire X-ray spectrum information following transmission through a subject. This study focused on the tissue composition identified using spectral information obtained by a new photon-counting detector. Normal breast tissue consists entirely of adipose and glandular tissues. However, it is very difficult to find tumor tissue in the region of glandular tissue via a conventional mammogram, especially in dense breast because the attenuation coefficients of glandular tissue and tumor tissue are very close. As a fundamental examination, we considered a simulation phantom and showed the difference between normal breast tissue and tumor tissue of various thicknesses in a three-dimensional (3D) scatter plot. We were able to discriminate between both types of tissues. In addition, there was a tendency for the distribution to depend on the thickness of the tumor tissue. Thinner tumor tissues were shown to be closer in appearance to normal breast tissue. This study also demonstrated that the difference between these tissues could be made obvious by using a CdTe series detector. We believe that this differentiation is important, and therefore, expect this technology to be applied to new tumor detection systems in the future.

In this work we present the MAMMOCARE prototype, a biopsy guided system based on PET. The system is composed by an examination table where the patient is situated in prone position, a PET detector and a biopsy device. The PET detector is composed by two rings. These rings can be separated mechanically in order to allow the needle insertion. The first acquisition is performed with the closed ring configuration in order to obtain a high quality image to locate the lesion. Then, the software calculates the optimum path for the biopsy and moves the biopsy and PET systems to the desired position. At this point, two compression pallets are used to hold the breast. Then, the PET system opens and the biopsy procedure starts. The images are obtained at several steps to ensure the correct location of the needle during the procedure. The performance of the system is evaluated measuring the spatial resolution and sensitivity according the NEMA standard. The uniformity of the reconstructed images is also estimated. The radial resolution is 1.62mm in the center of the FOV and 3.45mm at 50mm off the center in the radial direction using the closed configuration. In the open configuration the resolution reaches 1.85mm at center and 3.65mm at 50mm. The sensitivity using an energy window of 250keV-750keV is 3.6% for the closed configuration and 2.5% for the open configuration. The uniformity measured in the center of the FOV is 14% and 18% for the closed and open configurations respectively.

Breast cancer is one of the most common malignancies in women. For years, mammography has been used as the gold standard for localizing breast cancer, despite its limitation in determining cancer composition. Therefore, the purpose of this simulation study is to confirm the feasibility of obtaining tumor composition using dual energy digital mammography. To generate X-ray sources for dual energy mammography, 26 kVp and 39 kVp voltages were generated for low and high energy beams, respectively. Additionally, the energy subtraction and inverse mapping functions were applied to provide compositional images. The resultant images showed that the breast composition obtained by the inverse mapping function with cubic fitting achieved the highest accuracy and least noise. Furthermore, breast density analysis with cubic fitting showed less than 10% error compare to true values. In conclusion, this study demonstrated the feasibility of creating individual compositional images and capability of analyzing breast density effectively.

The photon counting detector with energy discrimination capabilities provides the spectral information and energy of each photon with single exposure. The energy-resolved photon counting detector makes it possible to improve the visualization of contrast agent by selecting the appropriate energy window. In this study, we simulated the photon counting spectral mammography system using a Monte Carlo method and compared three contrast enhancement methods (K-edge imaging, projection-based energy weighting imaging, and dual energy subtraction imaging). For the quantitative comparison, we used the homogeneous cylindrical breast phantom as a reference and the heterogeneous XCAT breast phantom. To evaluate the K-edge imaging methods, we obtained images by increasing the energy window width based on K-edge absorption energy of iodine. The iodine which has the K-edge discontinuity in the attenuation coefficient curve can be separated from the background. The projection-based energy weighting factor was defined as the difference in the transmissions between the contrast agent and the background. Each weighting factor as a function of photon energy was calculated and applied to the each energy bin. For the dual energy subtraction imaging, we acquired two images with below and above the iodine K-edge energy using single exposure. To suppress the breast tissue in high energy images, the weighting factor was applied as the ratio of the linear attenuation coefficients of the breast tissue at high and low energies. Our results demonstrated the CNR improvement of the K-edge imaging was the highest among the three methods. These imaging techniques based on the energy-resolved photon counting detector improved image quality with the spectral information.

This work investigated the effect of the grid-less acquisition mode with scatter correction software developed by Siemens Healthcare (PRIME mode) on image quality and mean glandular dose (MGD) in a comparative study against a standard mammography system with grid. Image quality was technically quantified with contrast-detail (c-d) analysis and by calculating detectability indices (d’) using a non-prewhitening with eye filter model observer (NPWE). MGD was estimated technically using slabs of PMMA and clinically on a set of 11439 patient images. The c-d analysis gave similar results for all mammographic systems examined, although the d’ values were slightly lower for the system with PRIME mode when compared to the same system in standard mode (-2.8% to -5.7%, depending on the PMMA thickness). The MGD values corresponding to the PMMA measurements with automatic exposure control indicated a dose reduction from 11.0% to 20.8% for the system with PRIME mode compared to the same system without PRIME mode. The largest dose reductions corresponded to the thinnest PMMA thicknesses. The results from the clinical dosimetry study showed an overall population-averaged dose reduction of 11.6% (up to 27.7% for thinner breasts) for PRIME mode compared to standard mode for breast thicknesses from 20 to 69 mm. These technical image quality measures were then supported using a clinically oriented study whereby simulated clusters of microcalcifications and masses were inserted into patient images and read by radiologists in an AFROC study to quantify their detectability. In line with the technical investigation, no significant difference was found between the two imaging modes (p-value 0.95).

This study aims to analyze the agreement between the mean glandular dose estimated by the mammography unit (organ dose) and mean glandular dose calculated using Dance et al published method (calculated dose). Anonymised digital mammograms from 50 BreastScreen NSW centers were downloaded and exposure information required for the calculation of dose was extracted from the DICOM header along with the organ dose estimated by the system. Data from quality assurance annual tests for the included centers were collected and used to calculate the mean glandular dose for each mammogram. Bland-Altman analysis and a two-tailed paired t-test were used to study the agreement between calculated and organ dose and the significance of any differences. A total of 27,869 dose points from 40 centers were included in the study, mean calculated dose and mean organ dose (± standard deviation) were 1.47 (±0.66) and 1.38 (±0.56) mGy respectively. A statistically significant 0.09 mGy bias (t = 69.25; p<0.0001) with 95% limits of agreement between calculated and organ doses ranging from -0.34 and 0.52 were shown by Bland-Altman analysis, which indicates a small yet highly significant difference between the two means. The use of organ dose for dose audits is done at the risk of over or underestimating the calculated dose, hence, further work is needed to identify the causal agents for differences between organ and calculated doses and to generate a correction factor for organ dose.

The scatter effect on detective quantum efficiency (DQE) of digital mammography is investigated using the cascaded-systems model. The cascaded-systems model includes a scatter-reduction device as a binomial selection stage. Quantum-noise-limited operation approximates the system DQE into the multiplication form of the scatter-reduction device DQE and the conventional detector DQE. The developed DQE model is validated in comparisons with the measured results using a CMOS flat-panel detector under scatter environments. For various scatter-reduction devices, the slot-scan method shows the best scatter-cleanup performance in terms of DQE, and the scatter-cleanup performance of the conventional one-dimensional grid is rather worse than the air gap. The developed model can also be applied to general radiography and will be very useful for a better design of imaging chain.

Screening for breast cancer using mammography has been very successful in the effort to reduce breast cancer mortality, and its use has largely resulted in the 30% reduction in breast cancer mortality observed since 1990 [1]. However, diagnostic mammography remains an area of breast imaging that is in great need for improvement. One imaging modality proposed for improving the accuracy of diagnostic workup is iodinated contrast-enhanced breast CT [2]. In this study, a mathematical framework is used to evaluate optimal exposure techniques for contrast-enhanced breast CT. The ideal observer signal-to-noise ratio (i.e., d’) figure-of-merit is used to provide a task performance based assessment of optimal acquisition parameters under the assumptions of a linear, shift-invariant imaging system. A parallel-cascade model was used to estimate signal and noise propagation through the detector, and a realistic lesion model with iodine uptake was embedded into a structured breast background. Ideal observer performance was investigated across kVp settings, filter materials, and filter thickness. Results indicated many kVp spectra/filter combinations can improve performance over currently used x-ray spectra.

Anthropomorphic software breast phantoms have been utilized for preclinical quantitative validation of breast imaging systems. Efficacy of the simulation-based validation depends on the realism of phantom images. Anatomical measurements of the breast tissue, such as the size and distribution of adipose compartments or the thickness of Cooper’s ligaments, are essential for the realistic simulation of breast anatomy. Such measurements are, however, not readily available in the literature. In this study, we assessed the statistics of adipose compartments as visualized in CT images of a total mastectomy specimen. The specimen was preserved in formalin, and imaged using a standard body CT protocol and high X-ray dose. A human operator manually segmented adipose compartments in reconstructed CT images using ITK-SNAP software, and calculated the volume of each compartment. In addition, the time needed for the manual segmentation and the operator’s confidence were recorded. The average volume, standard deviation, and the probability distribution of compartment volumes were estimated from 205 segmented adipose compartments. We also estimated the potential correlation between the segmentation time, operator’s confidence, and compartment volume. The statistical tests indicated that the estimated compartment volumes do not follow the normal distribution. The compartment volumes are found to be correlated with the segmentation time; no significant correlation between the volume and the operator confidence. The performed study is limited by the mastectomy specimen position. The analysis of compartment volumes will better inform development of more realistic breast anatomy simulation.

In order to improve efficacy of screening mammography, in recent years, we have been investigating the feasibility of applying a resonance-frequency based electrical impedance spectroscopy (REIS) technology to noninvasively detect breast abnormalities that may lead to the development of cancer in the near-term. Despite promising study-results, we found that REIS suffered from relatively poor reproducibility due to perturbations in electrode placement, contact pressure variation on the breast, as well as variation of the resonating inductor. To overcome this limitation, in this study, we propose and analyze a new paradigm of Dielectric Relaxation Spectroscopy (DRS) that measures polarization-lag of dielectric signals in breast-capacitance when excited by the pulses or sine waves. Unlike conventional DRS that operates using the signals at very high frequencies (GHz) to examine changes in polarization, our new method detects and characterizes the dielectric properties of tissue at low frequencies (≤10 MHz) due to the advent of inexpensive oscillators that are accurate to 1 pico-second (used in GPS receivers) as well as measurement of amplitudes of 1 ppm or better. From theoretical analysis, we have proved that the sensitivity of new DRS in detecting permittivity of water increased by ≥80 times as compared to conventional DRS, which operates at frequencies around 4GHz. By analyzing and comparing the relationship between the new DRS and REIS, we found that this DRS has potential advantages in enhancing repeatability from various readings, including temperature-insensitive detection, and yielding higher resolution or sensitivity (up to 100 Femtofarads).

Anthropomorphic software breast phantoms have been introduced as a tool for quantitative validation of breast imaging systems. Efficacy of the validation results depends on the realism of phantom images. The recursive partitioning algorithm based upon the octree simulation has been demonstrated as versatile and capable of efficiently generating large number of phantoms to support virtual clinical trials of breast imaging. Previously, we have observed specific artifacts, (here labeled “dents”) on the boundaries of simulated Cooper’s ligaments. In this work, we have demonstrated that these “dents” result from the approximate determination of the closest simulated ligament to an examined subvolume (i.e., octree node) of the phantom. We propose a modification of the algorithm that determines the closest ligament by considering a pre-specified number of neighboring ligaments selected based upon the functions that govern the shape of ligaments simulated in the subvolume. We have qualitatively and quantitatively demonstrated that the modified algorithm can lead to elimination or reduction of dent artifacts in software phantoms. In a proof-of concept example, we simulated a 450 ml phantom with 333 compartments at 100 micrometer resolution. After the proposed modification, we corrected 148,105 dents, with an average size of 5.27 voxels (5.27nl). We have also qualitatively analyzed the corresponding improvement in the appearance of simulated mammographic images. The proposed algorithm leads to reduction of linear and star-like artifacts in simulated phantom projections, which can be attributed to dents. Analysis of a larger number of phantoms is ongoing.

Purpose To evaluate the radiation dose derived from digital mammography (DM) and digital breast tomosynthesis (DBT) at different tube current-exposure time product (mAs) and at 6 phantom thicknesses from 10 to 60 mm. Materials and Methods A total of 240 DM and DBT cranio-caudal (CC) phantom images were acquired at each thickness and at four exposure levels (the baseline mAs, 50%, 25% and 12.5% the baseline mAs). The incident Air Kerma (K) at the surface of the phantoms was measured using a solid state dosimeter. Mean Glandular Doses (MGD) were calculated for both modalities (DM and DBT). Results DBT dose was greater than that of DM for all mAs at each phantom thickness. For a breast thickness of 50 mm (close to average sized breast), the dose for DBT (2.32 mGy) was 13% higher than that for DM (2.05 mGy). The results also show that the difference in MGD between DM and DBT was less for the thicker compared with the thinner phantom, this difference being approximately a factor of 2.58 at 10 mm compared with a factor of 1.08 at 60 mm. While the MGD increased with increasing phantom thickness for both modalities, the dose increase with DBT was less than for DM, with the difference between 10 and 60 mm being a factor of 7 for DM and 3 for DBT. Conclusion The radiation dose from DBT was higher than that of DM and the difference in dose between DM and DBT decreases as phantom thickness increases.

Our earlier work has demonstrated that the data consistency condition can be used as a criterion for scatter kernel optimization in deconvolution methods in a full-fan mode cone-beam CT [1]. However, this scheme cannot be directly applied to CBCT system with an offset detector (half-fan mode) because of transverse data truncation in projections. In this study, we proposed a modified scheme of the scatter kernel optimization method that can be used in a half-fan mode cone-beam CT, and have successfully shown its feasibility. Using the first-reconstructed volume image from half-fan projection data, we acquired full-fan projection data by forward projection synthesis. The synthesized full-fan projections were partly used to fill the truncated regions in the half-fan data. By doing so, we were able to utilize the existing data consistency-driven scatter kernel optimization method. The proposed method was validated by a simulation study using the XCAT numerical phantom and also by an experimental study using the ACS head phantom.

A high spatial resolution iterative reconstruction algorithm is proposed for a half cone beam CT (HCBCT) geometry. The proposed algorithm improves spatial resolution by explicitly accounting for image blurriness caused by different factors, such as extended X-ray source and detector response. The blurring kernel is estimated using the MTF slice of the Catphan phantom. The proposed algorithm is specifically optimized for the new Leksell Gamma Knife Icon (Elekta AB, Stockholm, Sweden) which incorporates the HCBCT geometry to accommodate the existing treatment couch while covering down to the base-of-skull in the reconstructed field-of-view. Image reconstruction involves a Fourier-based scaling simultaneous algebraic reconstruction technique (SART) coupled with total variation (TV) minimization and non-local mean denoising, solved using a split Bregman separation technique that splits the reconstruction problem into a gradient based updating step and a TV-based deconvolution algorithm. This formulation preserves edges and reduces the staircase effect caused by regular TV-penalized iterative algorithms. Our experiments indicate that our proposed method outperforms the conventional filtered back projection and TV penalized SART methods in terms of line pair resolution and retains the favorable properties of the standard TV-penalized reconstruction.

Cone-beam computed tomography (CBCT) has widely been used and studied in both medical imaging and radiation therapy. The aim of this study was to evaluate our newly developed CBCT system by implementing a distance-driven system modeling technique in order to produce excellent and accurate cross-sectional images. For the purpose of comparing the performance of the distance-driven methods, we also performed pixel-driven and ray-driven techniques when conducting forward- and back-projection schemes. We conducted the Feldkamp-Davis-Kress (FDK) algorithm and simultaneous algebraic reconstruction technique (SART) to retrieve a volumetric information of scanned chest phantom. The results indicated that contrast-to-noise (CNR) of the reconstructed images by using FDK and SART showed 8.02 and 15.78 for distance-driven, whereas 4.02 and 5.16 for pixel-driven scheme and 7.81 and 13.01 for ray-driven scheme, respectively. This could demonstrate that distance-driven method described more closely the chest phantom compared to pixel- and ray-driven. However, both elapsed time for modeling a system matrix and reconstruction time took longer time when performing the distance-driven scheme. Therefore, future works will be directed toward reducing computational time to acceptable limits for real applications.

The aim of this work is to develop a method to calculate lens dose for fluoroscopically-guided neuro-interventional procedures and for CBCT scans of the head. EGSnrc Monte Carlo software is used to determine the dose to the lens of the eye for the projection geometry and exposure parameters used in these procedures. This information is provided by a digital CAN bus on the Toshiba Infinix C-Arm system which is saved in a log file by the real-time skin-dose tracking system (DTS) we previously developed. The x-ray beam spectra on this machine were simulated using BEAMnrc. These spectra were compared to those determined by SpekCalc and validated through measured percent-depth-dose (PDD) curves and half-value-layer (HVL) measurements. We simulated CBCT procedures in DOSXYZnrc for a CTDI head phantom and compared the surface dose distribution with that measured with Gafchromic film, and also for an SK150 head phantom and compared the lens dose with that measured with an ionization chamber. Both methods demonstrated good agreement. Organ dose calculated for a simulated neuro-interventional-procedure using DOSXYZnrc with the Zubal CT voxel phantom agreed within 10% with that calculated by PCXMC code for most organs. To calculate the lens dose in a neuro-interventional procedure, we developed a library of normalized lens dose values for different projection angles and kVp’s. The total lens dose is then calculated by summing the values over all beam projections and can be included on the DTS report at the end of the procedure.

Modern CT image reconstruction algorithms rely on projection and back-projection operations to refine an image estimate in iterative image reconstruction. A widely-used state-of-the-art technique is distance-driven projection and back-projection. While the distance-driven technique yields superior image quality in iterative algorithms, it is a computationally demanding process. This has a detrimental effect on the relevance of the algorithms in clinical settings. A few methods have been proposed for enhancing the distance-driven technique in order to take advantage of modern computer hardware. This paper explores a two-dimensional extension of the branchless method proposed by Samit Basu and Bruno De Man. The extension of the branchless method is named “pre-integration” because it achieves a significant performance boost by integrating the data before the projection and back-projection operations. It was written with Nvidia’s CUDA platform and carefully designed for massively parallel GPUs. The performance and the image quality of the pre-integration method were analyzed. Both projection and back-projection are significantly faster with preintegration. The image quality was analyzed using cone beam image reconstruction algorithms within Jeffrey Fessler’s Image Reconstruction Toolbox. Images produced from regularized, iterative image reconstruction algorithms using the pre-integration method show no significant impact to image quality.

The skin-dose tracking system (DTS) provides a color-coded illustration of the cumulative skin-dose distribution on a closely-matching 3D graphic of the patient during fluoroscopic interventions in real-time for immediate feedback to the interventionist. The skin-dose tracking utility of DTS has been extended to include cone-beam computed tomography (CBCT) of neurointerventions. While the DTS was developed to track the entrance skin dose including backscatter, a significant part of the dose in CBCT is contributed by exit primary radiation and scatter due to the many overlapping projections during the rotational scan. The variation of backscatter inside and outside the collimated beam was measured with radiochromic film and a curve was fit to obtain a scatter spread function that could be applied in the DTS. Likewise, the exit dose distribution was measured with radiochromic film for a single projection and a correction factor was determined as a function of path length through the head. Both of these sources of skin dose are added for every projection in the CBCT scan to obtain a total dose mapping over the patient graphic. Results show the backscatter to follow a sigmoidal falloff near the edge of the beam, extending outside the beam as far as 8 cm. The exit dose measured for a cylindrical CTDI phantom was nearly 10 % of the entrance peak skin dose for the central ray. The dose mapping performed by the DTS for a CBCT scan was compared to that measured with radiochromic film and a CTDI-head phantom with good agreement.

Intracranial hemorrhage (ICH) is associated with pathologies such as hemorrhagic stroke and traumatic brain injury. Multi-detector CT is the current front-line imaging modality for detecting ICH (fresh blood contrast 40-80 HU, down to 1 mm). Flat-panel detector (FPD) cone-beam CT (CBCT) offers a potential alternative with a smaller scanner footprint, greater portability, and lower cost potentially well suited to deployment at the point of care outside standard diagnostic radiology and emergency room settings. Previous studies have suggested reliable detection of ICH down to 3 mm in CBCT using high-fidelity artifact correction and penalized weighted least-squared (PWLS) image reconstruction with a post-artifact-correction noise model. However, ICH reconstructed by traditional image regularization exhibits nonuniform spatial resolution and noise due to interaction between the statistical weights and regularization, which potentially degrades the detectability of ICH. In this work, we propose three regularization methods designed to overcome these challenges. The first two compute spatially varying certainty for uniform spatial resolution and noise, respectively. The third computes spatially varying regularization strength to achieve uniform "detectability," combining both spatial resolution and noise in a manner analogous to a delta-function detection task. Experiments were conducted on a CBCT test-bench, and image quality was evaluated for simulated ICH in different regions of an anthropomorphic head. The first two methods improved the uniformity in spatial resolution and noise compared to traditional regularization. The third exhibited the highest uniformity in detectability among all methods and best overall image quality. The proposed regularization provides a valuable means to achieve uniform image quality in CBCT of ICH and is being incorporated in a CBCT prototype for ICH imaging.

Three-dimensional cone-beam (CB) imaging using a multi-axis floor-mounted (or ceiling-mounted) C-arm system has become an important tool in interventional radiology. This success motivates new developments to improve image quality. One direction in which advancement is sought is the data acquisition geometry and related CB artifacts. Currently, data acquisition is performed using the circular short-scan trajectory, which yields limited axial coverage and also provides incomplete data for accurate reconstruction. To improve the image quality, as well as to increase the coverage in the longitudinal direction of the patient, we recently introduced the ellipse- line-ellipse trajectory and showed that this trajectory provides full R-line coverage within the field-of-view, which is a key property for accurate reconstruction from truncated data. An R-line is any segment of line that connects two source positions. Here, we examine how the application of asymmetrical variations to the definition of the ELE trajectory impacts the R-line coverage. This question is significant to understand how much flexibility can be used in the implementation of the ELE trajectory, particularly to adapt the scan to patient anatomy and imaging task of interest. Two types of asymmetrical variations, called axial and angular variations, are investigated.

Purpose: Scatter errors are detrimental to cone-beam breast CT (CBBCT) accuracy and obscure the visibility of calcifications and soft-tissue lesions. In this work, we propose practical yet effective scatter correction for CBBCT using a library-based method and investigate its feasibility via small-group patient studies. Method: Based on a simplified breast model with varying breast sizes, we generate a scatter library using Monte-Carlo (MC) simulation. Breasts are approximated as semi-ellipsoids with homogeneous glandular/adipose tissue mixture. On each patient CBBCT projection dataset, an initial estimate of scatter distribution is selected from the pre-computed scatter library by measuring the corresponding breast size on raw projections and the glandular fraction on a first-pass CBBCT reconstruction. Then the selected scatter distribution is modified by estimating the spatial translation of the breast between MC simulation and the clinical scan. Scatter correction is finally performed by subtracting the estimated scatter from raw projections. Results: On two sets of clinical patient CBBCT data with different breast sizes, the proposed method effectively reduces cupping artifact and improves the image contrast by an average factor of 2, with an efficient processing time of 200ms per conebeam projection. Conclusion: Compared with existing scatter correction approaches on CBBCT, the proposed library-based method is clinically advantageous in that it requires no additional scans or hardware modifications. As the MC simulations are pre-computed, our method achieves a high computational efficiency on each patient dataset. The library-based method has shown great promise as a practical tool for effective scatter correction on clinical CBBCT.

This paper is about the ring artifacts removal method in cone beam CT. Cone beam CT images often suffer from disturbance of ring artifacts which caused by the non-uniform responses of the elements in detectors. Conventional ring artifacts removal methods focus on the correlation of the elements and the ring artifacts’ structural characteristics in either sinogram domain or cross-section image. The challenge in the conventional methods is how to distinguish the artifacts from the intrinsic structures; hence they often give rise to the blurred image results due to over processing. In this paper, we investigate the characteristics of the ring artifacts in spatial space, different from the continuous essence of 3D texture feature of the scanned objects, the ring artifacts are displayed discontinuously in spatial space, specifically along z-axis. Thus we can easily recognize the ring artifacts in spatial space than in cross-section. As a result, we choose dictionary representation for ring artifacts removal due to its high sensitivity to structural information. We verified our theory both in spatial space and coronal-section, the experimental results demonstrate that our methods can remove the artifacts efficiently while maintaining image details.

X-ray scatter poses a significant limitation to image quality in cone-beam CT (CBCT), as well as beam hardening, resulting in image artifacts, contrast reduction, and lack of CT number accuracy. Meanwhile the x-ray radiation dose is also non-ignorable. Considerable scatter or beam hardening correction methods have been developed, independently, and rarely combined with low-dose CT reconstruction. In this paper, we combine scatter suppression with beam hardening correction for sparse-view CT reconstruction to improve CT image quality and reduce CT radiation. Firstly, scatter was measured, estimated, and removed using measurement-based methods, assuming that signal in the lead blocker shadow is only attributable to x-ray scatter. Secondly, beam hardening was modeled by estimating an equivalent attenuation coefficient at the effective energy, which was integrated into the forward projector of the algebraic reconstruction technique (ART). Finally, the compressed sensing (CS) iterative reconstruction is carried out for sparse-view CT reconstruction to reduce the CT radiation. Preliminary Monte Carlo simulated experiments indicate that with only about 25% of conventional dose, our method reduces the magnitude of cupping artifact by a factor of 6.1, increases the contrast by a factor of 1.4 and the CNR by a factor of 15. The proposed method could provide good reconstructed image from a few view projections, with effective suppression of artifacts caused by scatter and beam hardening, as well as reducing the radiation dose. With this proposed framework and modeling, it may provide a new way for low-dose CT imaging.

Metal in CT-imaged objects drastically reduces the quality of these images due to the severe artifacts it can cause. Most metal artifacts reduction (MAR) algorithms consider the metal-affected sinogram portions as the corrupted data and replace them via sophisticated interpolation methods. While these schemes are successful in removing the metal artifacts, they fail to recover some of the edge information. To address these problems, the frequency shift metal artifact reduction algorithm (FSMAR) was recently proposed. It exploits the information hidden in the uncorrected image and combines the high frequency (edge) components of the uncorrected image with the low frequency components of the corrected image. Although this can effectively transfer the edge information of the uncorrected image, it also introduces some unwanted artifacts. The essential problem of these algorithms is that they lack the capability of detecting the artifacts and as a result cannot discriminate between desired and undesired edges. We propose a scheme that does better in these respects. Our Metal Artifact Detection and Reduction (MADR) scheme constructs a weight map which stores whether a pixel in the uncorrected image belongs to an artifact region or a non-artifact region. This weight matrix is optimal in the Linear Minimum Mean Square Sense (LMMSE). Our results demonstrate that MADR outperforms the existing algorithms and ensures that the anatomical structures close to metal implants are better preserved.

In computed tomography (CT), metal implants increase the inconsistencies between the measured data and the linear assumption made by the analytical CT reconstruction algorithm. The inconsistencies appear in the form of dark and bright bands and streaks in the reconstructed image, collectively called metal artifacts. The standard method for metal artifact reduction (MAR) replaces the inconsistent data with the interpolated data. However, sinogram interpolation not only introduces new artifacts but it also suffers from the loss of detail near the implanted metals. With the help of a prior image that is usually estimated from the metal artifact-degraded image via computer vision techniques, improvements are feasible but still no MAR method exists that is widely accepted and utilized. We propose a technique that utilizes a prior image from a CT scan taken of the patient before implanting the metal objects. Hence there is a sufficient amount of structural similarity to cover the loss of detail around the metal implants. Using the prior scan and a segmentation or model of the metal implant our method then replaces sinogram interpolation with ray profile matching and estimation which yields much more reliable data estimates for the affected sinogram regions. As preliminary work, we built a new MAR framework on fan-beam geometry and tested it to remove simulated metal artifacts on a thorax phantom. The comparison with two representative sinogram correction based MAR methods shows very promising results.

Computed tomography (CT) has been used for medical purposes. However there are many artifacts at CT images and that makes distorted image. Ring artifact is caused by non-uniform sensitivity of detectors and makes ring shape artifact. Line-ratio method was proposed to solve the problem however there are some problem at specific case. Therefore we propose advanced method to correct ring artifact using transfer function. As a result, ring artifacts can be removed at more global cases. Simulation data shows the proposed method outperforms the conventional line-ratio method.

For myocardial perfusion CT exams, beam hardening (BH) artifacts may degrade the accuracy of myocardial perfusion defect detection. Meanwhile, cardiac motion may make BH process inconsistent, which makes conventional BH correction (BHC) methods ineffective. The aims of this study were to assess the severity of BH artifacts and motion artifacts and propose a projection-based iterative BHC method which has a potential to handle the motion-induced inconsistency better than conventional methods. In this study, four sets of forward projection data were first acquired using both cylindrical phantoms and cardiac images as objects: (1) with monochromatic x-rays without motion; (2) with polychromatic x-rays without motion; (3) with monochromatic x-rays with motion; and (4) with polychromatic x-rays with motion. From each dataset, images were reconstructed using filtered back projection; for datasets 2 and 4, one of the following BHC methods was also performed: (A) no BHC; (B) BHC that concerns water only; and (C) BHC that takes both water and iodine into account, which is an iterative method we developed in this work. Biases of images were quantified by the mean absolute difference (MAD). The MAD of images with BH artifacts alone (dataset 2, without BHC) was comparable or larger than that of images with motion artifacts alone (dataset 3): In the study of cardiac image, BH artifacts account for over 80% of the total artifacts. The use of BHC was effective: with dataset 4, MAD values were 170 HU with no BHC, 54 HU with water BHC, and 42 HU with the proposed BHC. Qualitative improvements in image quality were also noticeable in reconstructed images.

Accurate and artifact free reconstruction of tomographic images requires precise knowledge of the imaging system geometry. This work proposes a novel projection matrix (P-matrix) based calibration method to enable C-arm inverse geometry CT (IGCT). The method is evaluated for scanning-beam digital x-ray (SBDX), a C-arm mounted inverse geometry fluoroscopic technology. A helical configuration of fiducials is imaged at each gantry angle in a rotational acquisition. For each gantry angle, digital tomosynthesis is performed at multiple planes and a composite image analogous to a cone-beam projection is generated from the plane stack. The geometry of the C-arm, source array, and detector array is determined at each angle by constructing a parameterized 3D-to-2D projection matrix that minimizes the sum-of-squared deviations between measured and projected fiducial coordinates. Simulations were used to evaluate calibration performance with translations and rotations of the source and detector. In a geometry with 1 mm translation of the central ray relative to the axis-of-rotation and 1 degree yaw of the detector and source arrays, the maximum error in the recovered translational parameters was 0.4 mm and maximum error in the rotation parameter was 0.02 degrees. The relative rootmean- square error in a reconstruction of a numerical thorax phantom was 0.4% using the calibration method, versus 7.7% without calibration. Changes in source-detector-distance were the most challenging to estimate. Reconstruction of experimental SBDX data using the proposed method eliminated double contour artifacts present in a non-calibrated reconstruction. The proposed IGCT geometric calibration method reduces image artifacts when uncertainties exist in system geometry.

In this study, we propose to use patient-specific x-ray fluence control to reduce the radiation dose to sensitive organs while still achieving the desired image quality (IQ) in the region of interest (ROI). The mA modulation profile is optimized view by view, based on the sensitive organs and the ROI, which are obtained from an ultra-low-dose volumetric CT scout scan [1]. We use a clinical chest CT scan to demonstrate the feasibility of the proposed concept: the breast region is selected as the sensitive organ region while the cardiac region is selected as IQ ROI. Two groups of simulations are performed based on the clinical CT dataset: (1) a constant mA scan adjusted based on the patient attenuation (120 kVp, 300 mA), which serves as baseline; (2) an optimized scan with aggressive bowtie and ROI centering combined with patient-specific mA modulation. The results shows that the combination of the aggressive bowtie and the optimized mA modulation can result in 40% dose reduction in the breast region, while the IQ in the cardiac region is maintained. More generally, this paper demonstrates the general concept of using a 3D scout scan for optimal scan planning.

X-ray energy spectrum plays an essential role in imaging and related tasks. Due to the high photon flux of clinical CT scanners, most of the spectrum estimation methods are indirect and are usually suffered from various limitations. The recently proposed indirect transmission measurement-based method requires at least the segmentation of one material, which is insufficient for CT images of highly noisy and with artifacts. To combat for the bottleneck of spectrum estimation using segmented CT images, in this study, we develop a segmentation-free indirect transmission measurement based energy spectrum estimation method using dual-energy material decomposition. The general principle of the method is to compare polychromatic forward projection with raw projection to calibrate a set of unknown weights which are used to express the unknown spectrum together with a set of model spectra. After applying dual-energy material decomposition using high-and low-energy raw projection data, polychromatic forward projection is conducted on material-specific images. The unknown weights are then iteratively updated to minimize the difference between the raw projection and estimated projection. Both numerical simulations and experimental head phantom are used to evaluate the proposed method. The results indicate that the method provides accurate estimate of the spectrum and it may be attractive for dose calculations, artifacts correction and other clinical applications.

In previous studies of the noise power spectrum (NPS) of multi-detector CT (MDCT) systems, the image object was usually placed at the iso-center of the CT system; therefore, the bowtie filter had negligible impact on the shape of the two-dimensional (2D) NPS of MDCT. This work characterized the NPS of off-centered objects when a bowtie filter is present. It was found that the interplay between the bowtie filter and object position has significant impact on the rotational symmetry of the 2D NPS. Depending on the size of the bowtie filter, the degree of object off-centering, and the location of the region of interest (ROI) used for the NPS measurements, the symmetry of the 2D NPS can be classified as circular, dumbbell, and a peculiar cloverleaf symmetry. An anisotropic NPS corresponds to structured noise texture, which may directly influence the detection performance of certain low contrast detection tasks.

In recent years, there has been an increased interest in the development of dynamic bow-tie filters that are able to provide patient-specific x-ray beam shaping. We introduce the first physical prototype of a new adaptive bow-tie filter design based on the concept of “spatial exposure time modulation.” While most existing bow-tie filters operate by attenuating the radiation beam differently in different locations using partially attenuating objects, the presented filter shapes the radiation field using two movable completely radio-opaque collimators. The aperture and speed of the collimators is modulated in synchrony with the x-ray exposure to selectively block the radiation emitted to different parts of the object. This mode of operation does not allow the reproduction of every possible attenuation profile, but it can reproduce the profile of any object with an attenuation profile monotonically decreasing from the center to the periphery, such as an object with an elliptical cross section. Therefore, the new adaptive filter provides the same advantages as the currently existing static bow-tie filters, which are typically designed to work for a pre-determined cylindrical object at a fixed distance from the source, and provides the additional capability to adapt its performance at image acquisition time to better compensate for the actual diameter and location of the imaged object. A detailed description of the prototype filter, the implemented control methods, and a preliminary experimental validation of its performance are presented.

In thoracic CT, organ-based tube current modulation (OTCM) reduces breast dose by lowering the tube current in the 120° anterior dose reduction zone of patients. However, in practice the breasts usually expand to an angle larger than the dose reduction zone. This work aims to simulate a breast positioning technique (BPT) to constrain the breast tissue to within the dose reduction zone for OTCM and to evaluate the corresponding potential reduction in breast dose. Thirteen female anthropomorphic computational phantoms were studied (age range: 27-65 y.o., weight range: 52-105.8 kg). Each phantom was modeled in the supine position with and without application of the BPT. Attenuation-based tube current (ATCM, reference mA) was generated by a ray-tracing program, taking into account the patient attenuation change in the longitudinal and angular plane (CAREDose4D, Siemens Healthcare). OTCM was generated by reducing the mA to 20% between ± 60° anterior of the patient and increasing the mA in the remaining projections correspondingly (X-CARE, Siemens Healthcare) to maintain the mean tube current. Breast tissue dose was estimated using a validated Monte Carlo program for a commercial scanner (SOMATOM Definition Flash, Siemens Healthcare). Compared to standard tube current modulation, breast dose was significantly reduced using OTCM by 19.8±4.7%. With the BPT, breast dose was reduced by an additional 20.4±6.5% to 37.1±6.9%, using the same CTDIvol. BPT was more effective for phantoms simulating women with larger breasts with the average breast dose reduction of 30.2%, 39.2%, and 49.2% from OTCMBP to ATCM, using the same CTDIvol for phantoms with 0.5, 1.5, and 2.5 kg breasts, respectively. This study shows that a specially designed BPT improves the effectiveness of OTCM.

Dynamic myocardial perfusion computed tomography (MPCT) is a promising technique for diagnosis and risk stratification of coronary artery disease by assessing the myocardial perfusion hemodynamic maps (MPHM). Meanwhile, the repeated scanning of the same region results in a relatively large radiation dose to patients potentially. In this work, we present a robust MPCT deconvolution algorithm with adaptive-weighted tensor total variation regularization to estimate residue function accurately under the low-dose context, which is termed `MPD-AwTTV'. More specifically, the AwTTV regularization takes into account the anisotropic edge property of the MPCT images compared with the conventional total variation (TV) regularization, which can mitigate the drawbacks of TV regularization. Subsequently, an effective iterative algorithm was adopted to minimize the associative objective function. Experimental results on a modified XCAT phantom demonstrated that the present MPD-AwTTV algorithm outperforms and is superior to other existing deconvolution algorithms in terms of noise-induced artifacts suppression, edge details preservation and accurate MPHM estimation.

In computed tomography (CT), patient-specific organ dose can be estimated using pre-calculated organ dose conversion coefficients (organ dose normalized by CTDI_vol, h factor) database, taking into account patient size and scan coverage. The conversion coefficients have been previously estimated for routine body protocol classes, grouped by scan coverage, across an adult population for fixed tube current modulated CT. The coefficients, however, do not include the widely utilized tube current (mA) modulation scheme, which significantly impacts organ dose. This study aims to extend the h factors and the corresponding dose length product (DLP) to create effective dose conversion coefficients (k factor) database incorporating various tube current modulation strengths. Fifty-eight extended cardiac-torso (XCAT) phantoms were included in this study representing population anatomy variation in clinical practice. Four mA profiles, representing weak to strong mA dependency on body attenuation, were generated for each phantom and protocol class. A validated Monte Carlo program was used to simulate the organ dose. The organ dose and effective dose was further normalized by CTDI_vol and DLP to derive the h factors and k factors, respectively. The h factors and k factors were summarized in an exponential regression model as a function of body size. Such a population-based mathematical model can provide a comprehensive organ dose estimation given body size and CTDI_vol. The model was integrated into an iPhone app XCATdose version 2, enhancing the 1st version based upon fixed tube current modulation. With the organ dose calculator, physicists, physicians, and patients can conveniently estimate organ dose.

The purpose of this study was to substantiate the interdependency of image quality, radiation dose, and contrast material dose in CT towards the patient-specific optimization of the imaging protocols. The study deployed two phantom platforms. First, a variable sized phantom containing an iodinated insert was imaged on a representative CT scanner at multiple CTDI values. The contrast and noise were measured from the reconstructed images for each phantom diameter. Linearly related to iodine-concentration, contrast to noise ratio (CNR), was calculated for different iodine-concentration levels. Second, the analysis was extended to a recently developed suit of 58 virtual human models (5D-XCAT) with added contrast dynamics. Emulating a contrast-enhanced abdominal image procedure and targeting a peak-enhancement in aorta, each XCAT phantom was “imaged” using a CT simulation platform. 3D surfaces for each patient/size established the relationship between iodine-concentration, dose, and CNR. The Sensitivity of Ratio (SR), defined as ratio of change in iodine-concentration versus dose to yield a constant change in CNR was calculated and compared at high and low radiation dose for both phantom platforms. The results show that sensitivity of CNR to iodine concentration is larger at high radiation dose (up to 73%). The SR results were highly affected by radiation dose metric; CTDI or organ dose. Furthermore, results showed that the presence of contrast material could have a profound impact on optimization results (up to 45%).

Technology development in Computed Tomography (CT) is driven by clinical needs, for example the need for image quality sufficient for the clinical task, and the need to obtain the required image quality using the lowest possible radiation dose to the patient. One approach to manage dose without compromising image quality is to spatially vary the X-ray flux such that regions of high interest receive more radiation while regions of low interest or regions sensitive to radiation receive less dose. If the region of interest (ROI) is centered at the CT system’s axis of rotation, a simple stationary bowtie mounted between the X-ray tube and the patient is sufficient to reduce the X-ray flux outside the central region. If the ROI is off center, then a dynamic bowtie that can track the ROI as the gantry rotates is preferred. We experimentally demonstrated the dynamic bowtie using a design that is relatively simple, low cost, requires no auxiliary power supply, and can be retrofitted to an existing clinical CT scanner. We installed our prototype dynamic bowtie on a clinical CT scanner, and we scanned a phantom with a pre-selected off-center ROI. The dynamic bowtie reduced the X-ray intensity outside the targeted ROI tenfold. As a result, the reconstructed image shows significantly lower noise within the dynamic bowtie ROI compared to regions outside it. Our preliminary results suggest that a dynamic bowtie could be an effective solution for further reducing CT radiation dose.

We present a model for simulation of noisy X-ray computed tomography data sets. The model is made of two main components, a photon transport simulation component that generates the noiseless photon field incident on the detector, and a detector response model that takes as input the incident photon field parameters and given the X-ray source intensity and exposure time can generate noisy data sets, accordingly. The photon transport simulation component combines direct ray-tracing of polychromatic X-rays for calculation of transmitted data, with Monte Carlo simulation for calculation of the scattered-photon data. The Monte Carlo scatter simulation is accelerated by implementing particle splitting and importance sampling variance reduction techniques. The detector-incident photon field data are stored as energy expansion coefficients on a refined grid that covers the detector area. From these data the detector response model is able to generate noisy detector data realizations, by reconstituting the main parameters that describe each detector element response in statistical terms, including spatial correlations. The model is able to generate very fast, on the fly, CT data sets corresponding to different radiation doses, as well as detector response characteristics, facilitating data management in extensive optimization studies by reducing the computation time and storage space demands.

Currently computed tomography (CT) dosimetry relies on CT dose index (CTDI) and size specific dose estimates (SSDE). Organ dose is a better metric of radiation burden. However, organ dose estimation requires precise knowledge of organ locations. Regional imparted energy and dose can also be used to quantify radiation burden. Estimating the imparted energy from CT exams is beneficial in that it does not require precise estimates of the organ size or location. This work investigated an automated technique for retrospectively estimating the imparted energy from chest and abdominopelvic tube current modulated (TCM) CT exams. Monte Carlo simulations of chest and abdominopelvic TCM CT examinations across various tube potentials and TCM strengths were performed on 58 adult computational extended cardiac-torso (XCAT) phantoms to develop relationships between scanned mass and imparted energy normalized by dose length product (DLP). An automated algorithm for calculating the scanned patient volume was further developed using an open source mesh generation toolbox. The scanned patient volume was then used to estimate the scanned mass accounting for diverse density within the scan region. The scanned mass and DLP from the exam were used to estimate the imparted energy to the patient using the knowledgebase developed from the Monte Carlo simulations. Patientspecific imparted energy estimates were made from 20 chest and 20 abdominopelvic clinical CT exams. The average imparted energy was 274 ± 141 mJ and 681 ± 376 mJ for the chest and abdominopelvic exams, respectively. This method can be used to estimate the regional imparted energy and/or regional dose in chest and abdominopelvic TCM CT exams across clinical operations.

The authors introduce an algorithm to estimate the spatial dose distributions in computed tomography (CT) images. The algorithm calculates dose distributions due to the primary and scattered photons separately. The algorithm only requires the CT data set that includes the patient CT images and the scanner acquisition parameters. Otherwise the scanner acquisition parameters are extracted from the CT images. Using the developed algorithm, the dose distributions for head and chest phantoms are computed and the results show the excellent agreements with the dose distributions obtained using a commercial Monte Carlo code. The developed algorithm can be applied to a patient-specific CT dose estimation based on the CT data.

Evaluating spatial resolution is an essential work for cone-beam computed tomography (CBCT) manufacturers, prototype designers or equipment users. To investigate the cross-sectional spatial resolution for different transaxial slices with CBCT, the slanted edge technique with a 3D slanted edge phantom are proposed and implemented on a prototype cone-beam micro-CT. Three transaxial slices with different cone angles are under investigation. An over-sampled edge response function (ERF) is firstly generated from the intensity of the slightly tiled air to plastic edge in each row of the transaxial reconstruction image. Then the oversampled ESF is binned and smoothed. The derivative of the binned and smoothed ERF gives the line spread function (LSF). At last the presampled modulation transfer function (MTF) is calculated by taking the modulus of the Fourier transform of the LSF. The spatial resolution is quantified with the spatial frequencies at 10% MTF level and full-width-half-maximum (FWHM) value. The spatial frequencies at 10% of MTFs are 3.1±0.08mm^-1, 3.0±0.05mm^-1, and 3.2±0.04mm^-1 for the three transaxial slices at cone angles of 3.8°, 0°, and -3.8° respectively. The corresponding FWHMs are 252.8μm, 261.7μm and 253.6μm. Results indicate that cross-sectional spatial resolution has no much differences when transaxial slices being 3.8° away from z=0 plane for the prototype conebeam micro-CT.

An objective and standardized approach to assess image quality of Compute Tomography (CT) systems is required in a wide variety of imaging processes to identify CT systems appropriate for a given application. We present an overview of the framework we have developed to help standardize and to objectively assess CT image quality for different models of CT scanners used for security applications. Within this framework, we have developed methods to quantitatively measure metrics that should correlate with feature identification, detection accuracy and precision, and image registration capabilities of CT machines and to identify strengths and weaknesses in different CT imaging technologies in transportation security. To that end we have designed, developed and constructed phantoms that allow for systematic and repeatable measurements of roughly 88 image quality metrics, representing modulation transfer function, noise equivalent quanta, noise power spectra, slice sensitivity profiles, streak artifacts, CT number uniformity, CT number consistency, object length accuracy, CT number path length consistency, and object registration. Furthermore, we have developed a sophisticated MATLAB based image analysis tool kit to analyze CT generated images of phantoms and report these metrics in a format that is standardized across the considered models of CT scanners, allowing for comparative image quality analysis within a CT model or between different CT models. In addition, we have developed a modified sparse principal component analysis (SPCA) method to generate a modified set of PCA components as compared to the standard principal component analysis (PCA) with sparse loadings in conjunction with Hotelling T² statistical analysis method to compare, qualify, and detect faults in the tested systems.

We aim at improving low dose CT perfusion functional parameters maps and CT images quality, preserving quantitative information. In a dynamic CT perfusion dataset, each voxel is measured T times, where T is the number of acquired time points. In this sense, we can think about a voxel as a point in a T-dimensional space, where the coordinates of the voxels would be the values of its time attenuation curve (TAC). Starting from this idea, a k-means algorithm was designed to group voxels in K classes. A modified guided time-intensity profile similarity (gTIPS) filter was implemented and applied only for those voxels belonging to the same class. The approach was tested on a digital brain perfusion phantom as well as on clinical brain and body perfusion datasets, and compared to the original TIPS implementation. The TIPS filter showed the highest CNR improvement, but lowest spatial resolution. gTIPS proved to have the best combination of spatial resolution and CNR improvement for CT images, while k-gTIPS was superior to both gTIPS and TIPS in terms of perfusion maps image quality. We demonstrate k-means clustering analysis can be applied to denoise dynamic CT perfusion data and to improve functional maps. Beside the promising results, this approach has the major benefit of being independent from the perfusion model employed for functional parameters calculation. No similar approaches were found in literature.

A novel amorphous selenium (a-Se) direct detector with CMOS readout has been designed, and relative detector performance investigated. The detector features include a 25μm pixel pitch, and 1000μm thick a-Se layer operating at 10V/μm bias field. A simulated detector DQE was determined, and used in comparative calculations of the Relative Object Detectability (ROD) family of prewhitening matched-filter (PWMF) observer and non-pre-whitening matched filter (NPWMF) observer model metrics to gauge a-Se detector performance against existing high resolution micro-angiographic fluoroscopic (MAF) detectors and a standard flat panel detector (FPD). The PWMF-ROD or ROD metric compares two x-ray imaging detectors in their relative abilities in imaging a given object by taking the integral over spatial frequencies of the Fourier transform of the detector DQE weighted by an object function, divided by the comparable integral for a different detector. The generalized-ROD (G-ROD) metric incorporates clinically relevant parameters (focal- spot size, magnification, and scatter) to show the degradation in imaging performance for detectors that are part of an imaging chain. Preliminary ROD calculations using simulated spheres as the object predicted superior imaging performance by the a-Se detector as compared to existing detectors. New PWMF-G-ROD and NPWMF-G-ROD results still indicate better performance by the a-Se detector in an imaging chain over all sphere sizes for various focal spot sizes and magnifications, although a-Se performance advantages were degraded by focal spot blurring. Nevertheless, the a-Se technology has great potential to provide break- through abilities such as visualization of fine details including of neuro-vascular perforator vessels and of small vascular devices.

Pixelated photon counting detectors with energy discrimination capabilities are of increasing clinical interest for x-ray imaging. Such detectors, presently in clinical use for mammography and under development for breast tomosynthesis and spectral CT, usually employ in-pixel circuits based on crystalline silicon – a semiconductor material that is generally not well-suited for economic manufacture of large-area devices. One interesting alternative semiconductor is polycrystalline silicon (poly-Si), a thin-film technology capable of creating very large-area, monolithic devices. Similar to crystalline silicon, poly-Si allows implementation of the type of fast, complex, in-pixel circuitry required for photon counting – operating at processing speeds that are not possible with amorphous silicon (the material currently used for large-area, active matrix, flat-panel imagers). The pixel circuits of two-dimensional photon counting arrays are generally comprised of four stages: amplifier, comparator, clock generator and counter. The analog front-end (in particular, the amplifier) strongly influences performance and is therefore of interest to study. In this paper, the relationship between incident and output count rate of the analog front-end is explored under diagnostic imaging conditions for a promising poly-Si based design. The input to the amplifier is modeled in the time domain assuming a realistic input x-ray spectrum. Simulations of circuits based on poly-Si thin-film transistors are used to determine the resulting output count rate as a function of input count rate, energy discrimination threshold and operating conditions.

X-ray mammography is a crucial screening tool for early identification of breast cancer. However, the overlap of anatomical features present in projection images often complicates the task of correctly identifying suspicious masses. As a result, there has been increasing interest in acquisition of volumetric information through digital breast tomosynthesis (DBT) which, compared to mammography, offers the advantage of depth information. Since DBT requires acquisition of many projection images, it is desirable that the noise in each projection image be dominated by the statistical noise of the incident x-ray quanta and not by the additive noise of the imaging system (referred to as quantum-limited imaging) and that the cumulative dose be as low as possible (e.g., no more than for a mammogram). Unfortunately, the electronic noise (~2000 electrons) present in current DBT systems based on active matrix, flat-panel imagers (AMFPIs) is still relatively high compared with modest x-ray gain of the a-Se and CsI:Tl x-ray converters often used. To overcome the modest signal-to-noise ratio (SNR) limitations of current DBT systems, we have developed a large-area x-ray imaging detector with the combination of an extremely low noise (~20 electrons) active-pixel CMOS and a specially designed high resolution scintillator. The high sensitivity and low noise of such system provides better SNR by at least an order of magnitude than current state-of-art AMFPI systems and enables x-ray indirect-detection single photon counting (SPC) at mammographic energies with the potential of dose reduction.

Photon counting detectors (PCDs) have the potential to improve x-ray imaging, however they are still hindered by several performance limitations and high production cost. By using amorphous Selenium (a-Se) the cost of PCDs can be significantly reduced compared to crystalline materials and enable large area detector fabrication. To overcome the problem of low carrier mobility and low charge conversion gain in a-Se, we are developing a novel direct conversion a- Se field-Shaping multi-Well Avalanche Detector (SWAD). SWAD circumvents the charge transport limitation by using a Frisch grid built within the readout circuit, reducing charge collection time to ~200 ns. Field shaping permits depth independent avalanche gain in wells, resulting in total conversion gain that is comparable to Si and CdTe. In the present work we investigate the effects of charge sharing and energy loss to understand the inherent photon counting performance for SWAD at x-ray energies used in breast imaging applications (20-50keV). The energy deposition profile for each interacting x-ray was determined with Monte Carlo simulation. For the energy ranges we are interested in, photoelectric interaction dominates, with a k-fluorescence yield of approximately 60%. Using a monoenergetic 45 keV beam incident on a target pixel in 400um of a-Se, our results show that only 20.42 % and 22.4 % of primary interacting photons have kfluorescence emissions which escape the target pixel for 100um and 85um pixel sizes respectively, demonstrating SWAD’s potential for high spatial resolution applications.

Small pixel high resolution direct x-ray detectors have the advantage of higher spatial sampling and decreased blurring characteristic. The limiting factors for such systems becomes the degradation due to the focal spot size. One solution is a smaller focal spot; however, this can limit the output of the x-ray tube. Here a software solution of deconvolving with an estimated focal spot blur is presented. To simulate images from a direct detector affected with focal-spot blur, first a set of high-resolution stent images (FRED from Microvention, Inc., Tustin, CA) were acquired using a 75μm pixel size Dexela-Perkin-Elmer detector and frame averaged to reduce quantum noise. Then the averaged image was blurred with a known Gaussian blur. To add noise to the blurred image a flat-field image was multiplied with the blurred image. Both the ideal and the noisy-blurred images were then deconvolved with the known Gaussian function using either threshold-based inverse filtering or Weiner deconvolution. The blur in the ideal image was removed and the details were recovered successfully. However, the inverse filtering deconvolution process is extremely susceptible to noise. The Weiner deconvolution process was able to recover more of the details of the stent from the noisy-blurred image, but for noisier images, stent details are still lost in the recovery process.

The fixed pattern noise, which is due to the nonuniform amplifier gains and scintillator sensitivities, should be alleviated in radiography imaging and should have less influence on measuring the noise power spectrum (NPS) of the radiography detector. In order to reduce the influence, background trend removing methods, which are based on low-pass filtering, polynomial fitting, and subtracting the average image of the uniform exposure images, are traditionally employed in the literature. In terms of removing the fixed pattern noise, the subtraction method shows a good performance. However, the number of images to be averaged is practically finite and thus the noise contained in the average image contaminates the image difference and inflates the NPS curve. In this paper, an image formation model considering the nonuniform gain is constructed and two measuring methods, which are based on the subtraction and gain correction, respectively, are considered. In order to accurately measure a normalized NPS (NNPS) in the measuring methods, the number of images to be averaged is considered for NNPS compensations. For several flat-panel radiography detectors, the NNPS measurements are conducted and compared with conventional approaches, which have no compensation stages. Through experiments it is shown that the compensation can provide accurate NNPS measurements less influenced by the fixed pattern noise.

Quantum-limited-dose (QLD) and noise-equivalent-dose (NED) are performance metrics often used interchangeably. Although the metrics are related, they are not equivalent unless the treatment of electronic noise is carefully considered. These metrics are increasingly important to properly characterize the low-dose performance of flat panel detectors (FPDs). A system can be said to be quantum-limited when the Signal-to-noise-ratio (SNR) is proportional to the square-root of x-ray exposure. Recent experiments utilizing three methods to determine the quantum-limited dose range yielded inconsistent results. To investigate the deviation in results, generalized analytical equations are developed to model the image processing and analysis of each method. We test the generalized expression for both radiographic and fluoroscopic detectors. The resulting analysis shows that total noise content of the images processed by each method are inherently different based on their readout scheme. Finally, it will be shown that the NED is equivalent to the instrumentation-noise-equivalent-exposure (INEE) and furthermore that the NED is derived from the quantum-noise-only method of determining QLD. Future investigations will measure quantum-limited performance of radiographic panels with a modified readout scheme to allow for noise improvements similar to measurements performed with fluoroscopic detectors.

Dual-energy imaging method has been introduced to improve conspicuity of abnormalities in radiographs. The method typically uses the fast kilovoltage-switching approach, which acquires low and high-energy projections in successive x-ray exposures with the same detector. However, it is typically known that there exists an optimal detector thickness regarding specific imaging tasks or energies used. In this study, the dual-energy detectability has been theoretically addressed for various combinations of detector thicknesses for low and high-energy spectra using the cascaded-systems analysis. Cesium iodide (CsI) is accounted for the x-ray converter in the hypothetical detector. The simple prewhitening model shows that a larger CsI thickness (250 mg cm^-2 for example) would be preferred to the the typical CsI thickness of 200 mg cm^-2 for better detectability. On the other hand, the typical CsI thickness is acceptable for the prewhitening model considering human-eye filter. The theoretical strategy performed in this study will be useful for a better design of detectors for dual-energy imaging.

Flat panel detectors have become the standard technology in projection radiography. Further progress in detector technology will result in an improvement of MTF and DQE. The new detector (DX-D45C; Agfa; Mortsel/Belgium) is based on cesium-iodine crystals and has a change in the detector material and the readout electronics. The detector has a size of 30 cm x 24 cm and a pixel matrix of 2560 x 2048 with a pixel pitch of 124 μm. The system includes an automatic exposure detector, which enables the use of the detector without a connection to the x-ray generator. The physical properties of the detector were determined following IEC 62220-1-1 in a laboratory setting. The MTF showed an improvement compared to the previous version of cesium-iodine based flat-panel detectors. Thereby the DQE is also improved especially for the higher frequencies. The new detector showed an improvement in the physical properties compared to the previous versions. This enables a potential for further dose reductions in clinical imaging.

The detective quantum efficiency (DQE), expressed as a function of spatial frequency, describes the ability of an x-ray detector to produce high signal-to-noise ratio (SNR) images. While regulatory and scientific communities have used the DQE as a primary metric for optimizing detector design, the DQE is rarely used by end users to ensure high system performance is maintained. Of concern is that image quality varies across different systems for the same exposures with no current measures available to describe system performance. Therefore, here we conducted an initial DQE measurement survey of clinical x-ray systems using a DQE-testing instrument to identify their range of performance. Following laboratory validation, experiments revealed that the DQE of five different systems under the same exposure level (8.0 μGy) ranged from 0.36 to 0.75 at low spatial frequencies, and 0.02 to 0.4 at high spatial frequencies (3.5 cycles/mm). Furthermore, the DQE dropped substantially with decreasing detector exposure by a factor of up to 1.5x in the lowest spatial frequency, and a factor of 10x at 3.5 cycles/mm due to the effect of detector readout noise. It is concluded that DQE specifications in purchasing decisions, combined with periodic DQE testing, are important factors to ensure patients receive the health benefits of high-quality images for low x-ray exposures.

The actual meaning of the modulation-transfer function (MTF) and the noise-power spectrum (NPS) is ambiguous in dual-energy images obtained from the single-shot sandwich detector, and their properties for various detector design parameters are also being questioned. In this study, the authors regard the sandwich detector including weighted logarithmic subtraction operation as a single black-box detector, and measure the single-shot dual-energy MTF and NPS performances. Subtraction of two images obtained from the sub-detector layers, which have different thick x-ray converters (hence, different spatial-resolution performances), of the sandwich detector yields a band-pass filter characteristic of the MTF. On the other hand, the NPS is the weighted sum of each NPS obtained from the sub-detector layers. The MTF characteristic is reflected into the DQE, hence the DQE shows a similar band-pass filter characteristics. Therefore, the sandwich detector may lose the contrast performance for large-area objects, but it may emphasize the contrast performance for objects with importance at mid-frequency information.

Single-shot dual-energy sandwich detector can produce sharp images because of subtraction of images from two sub-detector layers, which have different thick x-ray converters, of the sandwich detector. Inspired by this observation, the authors have developed a microtomography system with the sandwich detector in pursuit of high-resolution bone-enhanced small-animal imaging. The preliminary results show that the bone-enhanced images reconstructed with the subtracted projection data are better in visibility of bone details than the conventionally reconstructed images. In addition, the bone-enhanced images obtained from the sandwich detector are relatively immune to the artifacts caused by photon starvation. The microtomography with the single-shot dual-energy sandwich detector will be useful for the high-resolution bone imaging.

We propose a new design of a stacked three-layer flat-panel x-ray detector for dual-energy (DE) imaging. Each layer consists of its own scintillator of individual thickness and an underlying thin-film-transistor-based flat-panel. Three images are obtained simultaneously in the detector during the same x-ray exposure, thereby eliminating any motion artifacts. The detector operation is two-fold: a conventional radiography image can be obtained by combining all three layers’ images, while a DE subtraction image can be obtained from the front and back layers’ images, where the middle layer acts as a mid-filter that helps achieve spectral separation. We proceed to optimize the detector parameters for two sample imaging tasks that could particularly benefit from this new detector by obtaining the best possible signal to noise ratio per root entrance exposure using well-established theoretical models adapted to fit our new design. These results are compared to a conventional DE temporal subtraction detector and a single-shot DE subtraction detector with a copper mid-filter, both of which underwent the same theoretical optimization. The findings are then validated using advanced Monte Carlo simulations for all optimized detector setups. Given the performance expected from initial results and the recent decrease in price for digital x-ray detectors, the simplicity of the three-layer stacked imager approach appears promising to usher in a new generation of multi-spectral digital x-ray diagnostics.

In energy-resolved CT, images are reconstructed independently at different energy levels, resulting in images with different qualities but the same structures. We propose a similarity-based non-local filtration method to extract structural information from these images for noise suppression. For each pixel, we calculate its similarity to other pixels based on CT number. The calculation is repeated on each image at different energy levels and similarity values are averaged to generate a similarity matrix. Noise suppression is achieved by multiplying the image vector by the similarity matrix. Multiple scans on a tabletop CT system are used to simulate 6-channel energy-resolved CT, with energies ranging from 75 to 125 kVp. Phantom studies show that the proposed method improves average contrast-to-noise ratio (CNR) of seven materials on the 75 kVp image by a factor of 22. Compared with averaging CT images for noise suppression, our method achieves a higher CNR and reduces the CT number error of iodine solutions from 16.5% to 3.5% and the overall image root of mean-square error (RMSE) from 3.58% to 0.93%. On the phantom with line-pair structures, our algorithm reduces noise standard deviation (STD) by a factor of 23 while maintaining 7 lp/cm spatial resolution. Additionally, anthropomorphic head phantom studies show noise STD reduction by a factor or 26 with no loss of spatial resolution. The noise suppression achieved by the similarity-based method is clinically attractive, especially for CNRs of iodine in contrast-enhanced CT.

Coronary artery calcification (CAC) as assessed with CT calcium score is the best biomarker of coronary artery disease. Dual energy x-ray provides an inexpensive, low radiation-dose alternative. A two shot system (GE Revolution-XRd) is used, raw images are processed with a custom algorithm, and a coronary calcium image (DECCI) is created, similar to the bone image, but optimized for CAC visualization, not lung visualization. In this report, we developed a physicsbased, digital-phantom containing heart, lung, CAC, spine, ribs, pulmonary artery, and adipose elements, examined effects on DECCI, suggested physics-inspired algorithms to improve CAC contrast, and evaluated the correlation between CT calcium scores and a proposed DE calcium score. In simulation experiment, Beam hardening from increasing adipose thickness (2cm to 8cm) reduced C_g by 19% and 27% in 120kVp and 60kVp images, but only reduced C_g by <7% in DECCI. If a pulmonary artery moves or pulsates with blood filling between exposures, it can give rise to a significantly confounding PA signal in DECCI similar in amplitude to CAC. Observations suggest modifications to DECCI processing, which can further improve CAC contrast by a factor of 2 in clinical exams. The DE score had the best correlation with "CT mass score" among three commonly used CT scores. Results suggest that DE x-ray is a promising tool for imaging and scoring CAC, and there still remains opportunity for further DECCI processing improvements.