KEYWORDS: Reconstruction algorithms, Image restoration, Model based design, Tissues, Diffuse optical tomography, Data modeling, Matrices, Algorithm development, Chromophores, Signal to noise ratio
Diffuse optical tomography (DOT) is a promising non-invasive optical imaging technology that can provide functional information of biological tissues. Since the diffused light undergoes multiple scattering in biological tissues, and the boundary measurements are limited, the inverse problem of DOT is ill-posed and ill-conditioned. To overcome these limitations, inverse problems in DOT are often mitigated using regularization techniques, which use data fitting and regularization terms to suppress the effects of measurement noise and modeling errors. Tikhonov regularization, utilizing the L2 norm as its regularization term, often leads to images that are excessively smooth. In recent years, with the continuous development of deep learning algorithms, many researchers have used Model-based deep learning methods for reconstruction. However, the reconstruction of DOT is solved on mesh, arising from a finite element method for inverse problems, it is difficult to use it directly for convolutional network. Therefore, we propose a model-based graph convolutional network (Model-GCN). Overall, Model-GCN achieves better image reconstruction results compared to Tikhonov, with lower absolute bias error (ABE). Specifically, for total hemoglobin (HbT) and water, the average reduction in ABE is 68.3% and 77.3%, respectively. Additionally, the peak signal-to-noise (PSNR) values are on average increased by 6.4dB and 7.0dB.
Diffuse Optical Tomography (DOT) is a promising noninvasive imaging method which quantifies optical parameters to achieve functional characteristics of biological tissues. Since boundary measurements are limited and the light propagation in biological tissues is highly diffusive, DOT image reconstruction is ill-posed and ill-conditioned. Traditional iteration-based regularization methods, such as Tikhonov regularization, tend to produce low-quality images with severe artifacts. Deep learning reconstruction methods based on convolutional neural networks may ignore the structure information of finite element mesh during feature extraction and cause the loss of spatial features. In order to overcome these problems, we propose a novel neural network based on graph convolution to improve the quality of reconstructed DOT images. Graph convolution can make full use of the spatial structure information of finite element mesh and deal with the potential correlation between optical features effectively. To verify the performance of the proposed algorithm, simulation experiments were performed. We also compared the proposed algorithm with Tikhonov regularization method and deep learning method based on convolutional neural network. Experimental results show that proposed algorithm obtains better reconstruction results in terms of ABE, MSE, PSNR and SSIM.
Diffuse optical tomography (DOT) is a promising non-invasive optical imaging technique that can provide functional information of biological tissues. Since diffuse light undergoes multiple scattering in biological tissues and boundary measurements are limited, DOT reconstruction is ill-posedness and ill-conditioned. To overcome these limitations, Tikhonov regularization is the most popular algorithm. Recently, deep learning based reconstruction methods have attracted increasing attention, and promising results have been reported. However, they lack generalization for unstructured physical model. Therefore, a model-base convolution neural network framework (Model-CNN) is developed. It composes of two layers: data consistency layer and depth layer, which increases the interpretability of the model. Its performance is evaluated with numerical simulations. Our results demonstrate that Model-CNN can get better reconstructed results than those obtained by Tikhonov Regularization in terms of ABE, MSE, and PSNR.
A new low-cost imaging system has been developed for MRI-guided Near-Infrared Spectral Tomography (MRI-NIRST) for breast cancer detection. In this new system, 8 flexible sensing strips with 4 source fibers and 4 photodetectors will be applied directly to the breast, from nipple towards chest-wall, to cover the entire breast. For each source illumination, data will be collected at 32 detector locations. The feasibility of this new imaging system was demonstrated using 3D simulation breast phantoms that created from real breast MR images. Our results indicated that total hemoglobin in a tumor smaller than 1 mm could be recovered with this system.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.