This conference presentation was prepared for SPIE BiOS, 2023.

This conference presentation was prepared for SPIE BiOS, 2023.

I cherish my time with Gabi when we both were at MIT. He was a postdoc in the George R. Harrison Spectroscopy Laboratory led by Prof. Michael S. Feld and I was a graduate student in the Laser Interferometer Gravitational-wave Laboratory (LIGO) led by Rainer Weiss (Nobel laureate in physics). We got to know each other, not through research meetings, but through weekly soccer activity in the physics department. Every week after soccer practice sessions, we exchanged research ideas and brainstormed how to combine our expertise, which led to the publication of an excellent joint paper on nanometer-resolved surface vibrometry of cells in Physical Review Letters. Through this process, we both learned the importance of open-mindedness for impactful collaborative research. In this talk, I will share this memory with the audience.

I will share my personal interactions with Prof. Gabriel Popescu (known as Gabi), who was a pioneer in Quantitative Phase Imaging (QPI). I will reflect my PhD years, learning QPI with Gabi. I will also share his wisdom and his passion in scientific research.

If I had not met Gabi as my tensor, I would not have continued my research career. I was fortunate to start my graduate research projects with Gabi. In 2005, Gabi was a postdoc and I was a fresh graduate student in George R. Harrison Spectroscopy Laboratory led by Prof. Michael S. Feld. My very first research project was the combination of QPI and fluorescence imaging, which was supervised by Gabi. I have learned physics, optics, and more importantly attitude towards research and life, from Gabi. We have discussed numerous interesting ideas over beer, which shaped my current research fields. In this talk, I will share the early research results I have worked with Gabi 15 years ago, and the latest research progress which was inspired and influenced from Gabi.

I will discuss our efforts of pushing the limit of computational phase tomography. I will discuss Intensity Diffraction Tomography (IDT) for quantitative 3D phase imaging. IDT can be easily implemented in a standard microscope equipped with a programmable light source, making it easily accessible to the biological research community. I will present both physical model and deep learning strategies for improving the imaging capabilities of IDT for handling complex 3D biological samples. Finally, I will present a novel technique of combining IDT and mid-infrared photothermal imaging to enable Bond-selective phase tomography.

Rapid identification of infectious pathogens can save lives and mitigate healthcare expenses. Yet the current turnaround time for microbial identification typically exceeds 24 hours, as the common methods require the cultivation of millions or more bacteria to detect the collective signal. In this study, we propose a hybrid framework of quantitative phase imaging and artificial neural network to facilitate rapid identification at an individual-cell level. Specifically, three-dimensional images of refractive index were acquired for individual bacteria, and an optimized artificial neural network determined the species based on the three-dimensional morphologies, securing 82.5% blind test accuracy at an individual-cell level.

With applications ranging from metabolomics to histopathology, quantitative phase microscopy (QPM) is a powerful label-free imaging modality. However, the speed of current QPM systems is limited by electronic hardware. To improve throughput further, here we propose differentiable optical-electronic quantitative phase microscopy (∂μ) that acquires images in a compressed form such that more information can be transferred beyond the electronic hardware bottleneck. The proposed microscopy uses optical feature extractors as image compressors. The resultant intensity distribution is then decompressed into QPM images by a reconstruction network. By optimizing optical-electronic parameters in an end-to-end manner, our method can improve the QPI throughput from Hours to Seconds (more than an order of magnitude).

A novel multi-modal label-free imaging system is proposed for histopathology, which provides uniformly reconstructed virtual-stained brightfield images and corresponding QPI images. The system was tested on urinal histopathology, to detect and segment glomerulus. From each modality, over 90% of IoU scores were obtained and accelerated performance was obtained through multi-modal learning. Briefly, histopathology quantification with label-free samples is a feasible method via the proposed novel system.

We report a novel few-shot transfer learning scheme based on a convolutional recurrent neural network architecture, which was used for holographic image reconstruction. Without sacrificing the hologram reconstruction accuracy and quality, this few-shot transfer learning scheme effectively reduced the number of trainable parameters during the transfer learning process by ~90% and improved the convergence speed by 2.5-fold over baseline models. This method can be applied to other deep learning-based computational microscopy and holographic imaging tasks, and facilitates the transfer learning of models to new types of samples with minimal training time and data.

Since its introduction, diffractive deep neural networks (D2NN) have been an emerging technology with many useful applications, such as 3D object recognition, saliency segmentation, and quantitative phase microscopy. However, fabricating D2NNs operating in the visible range has not been performed due to the complexities of fabricating nano-scale elements. Recent advancements such as Implosion Fabrication have made it possible to fabricate such networks using a discrete number of phase weights. We propose a quantization-aware training approach for D2NNs through modeling the quantization process in a differentiable manner using a sigmoid-based quantization function, facilitating the fabrication process. We also propose an efficient training schedule to guide the optimization process to converge to a better minima despite the limited number of quantization levels. Our method is simulated and validated for an all-optical quantitative phase microscopy task based on the phase D2NN.

Polychromatic digital holographic microscopy (P-DHM) has proven its capacity to provide quasi-coherent noise-free quantitative-phase images, allowing a high-quality visualization of cell structure. In this work we propose a fully automated hologram reconstruction methodology, including a fast-numerical approach for the correction of the defocusing resulting from both the axial chromatic aberrations as well as the fluctuations of the optomechanical elements. This methodology able to reconstruct a large number of holograms paves the way to develop a time-resolved P-DHM capable of non-invasively visualizing both the fine cell structure and dynamics.

Digital holographic microscopy (DHM), provides an extremely sensitive quantitative-phase signal (QPS), which is nevertheless affected by coherent noise. The recent development of polychromatic DHM (P-DHM) enables us to provide quasi-coherent-noise-free quantitative-phase images. The implementation of P-DHM remains, however, demanding. We propose a convolutional neural network architecture, using for the first time an experimental ground-truth dataset, performing the P-DHM denoising from conventional DHM images. The results highlight, a strong efficiency, fine subcellular structures are made visible without loss of QPS accuracy, an interest in comparison to state-of-the-art learning methods and the possibility of a more widespread use of the P-DHM.

This conference presentation was prepared for SPIE BiOS, 2023.

We report high-speed and highly sensitive quantitative phase microscopy (QPM) using dynamic speckle illumination (DSI). The DSI-QPM is used for real-time analyses of highly motile human spermatozoa. The DSI-QPM supports high-speed and high spatial phase sensitivity, that are crucial for imaging tail (nanoscale) of living spermatozoa during motion. The scalable FoV and high temporal coherence offered by DSI-QPM is harnessed for histopathology and marine biology. Further, by integrating the single molecule localization microscopy (SMLM) with QPM, nanoscale imaging and quantification in lateral (via SMLM) and axial (via QPM) directions was achieved on liver cells.

We present an acoustic-driven quantitative phase microscope for quantifying cell biomechanics at sub-cellular level. Longitudinal acoustic standing wave is established as mechanical loadings to deform cells. A reflection quantitative phase microscopy with high axial and lateral resolution is used to quantify the cell deformations. We evaluate cancerous and normal cells undergoing standing wave modulation, where both the time-lapse cell membrane fluctuations and whole-cell volumetric shape changes are measured. Our technique can achieve high throughput, label-free and non-invasive quantification of cell mechanics with high mechanical sensitivity over broad mechanical frequency response range and can potentially be extended to study 3D tissue mechanics with sub-cellular resolution.

Visualization of refractive index (RI) distribution of thick specimens is challenging due to multiple-scattering and sample-induced aberration caused by the inhomogeneity of the sample. Here, we propose a novel RI reconstruction method that overcoming these detrimental effects by partial reconstruction of RI distribution and computational wave backpropagation through the partial RI distribution. Exploiting this computational approach, we demonstrate the subcellular-resolution visualization of multicellular spheroid with a diameter of 140 µm. We further introduce the latest results on various cell-type spheroids with different morphology that demonstrate the generalizability of this approach, and discuss applicability to biomedical fields.

Fourier ptychography (FP) utilizes angle-varied illumination to achieve resolution improvement and quantitative phase imaging. In this talk, we present a compact microscope using an OLED screen as a programmable illumination for FP reconstruction. We discuss multiplexed reconstruction strategy using multi-pixel illuminations, and a stand-alone smartphone implementation of portable FPM.

Quantitative phase imaging (QPI) techniques require multiple measurements to obtain the refractive index (RI) distribution of a sample. Here, we present a single-shot RI imaging method using spectral multiplexing and optical transfer function reshaping. In the present method, we simultaneously measure three intensity images of a sample with three optimized illumination patterns. Deconvolution of the measured intensity images is then performed to obtain the RI distribution of the sample. As a proof-of-concept, we measured both microspheres and biological cells.

Optical imaging is a major tool in the basic sciences, and provides both morphological and molecular-specific imaging capabilities. Unfortunately, optical scattering is a major obstacle for high-resolution imaging because scattering scrambles object-specific information. I present some of our recent developments that utilize concepts from computational phase-retrieval and physics-based iterative solvers to image through scattering samples. The two main application spaces we target is 1) using optical-scattering to enable large space-bandwidth imaging, and 2) computationally decoding 3D tissue scattering.

We report diffractive optical networks to replace the digital image reconstruction schemes in quantitative phase imaging (QPI) systems with a series of passive diffractive surfaces that perform phase recovery all-optically, completing the reconstruction of the QPI signal as the light is diffracted through thin structured layers. A diffractive QPI network all-optically maps the input phase information into an intensity distribution that reveals the QPI signal at the speed of light. With their compact footprint, diffractive QPI networks offer ultra-fast all-optical computation and low-power operation, demonstrating a new phase imaging concept that can be especially transformative for on-chip microscopy and sensing.

Nanoscale nuclear architecture mapping (nanoNAM) is a label free imaging method based on the Fourier phase of Fourier-domain optical coherence tomography. It is capable of capturing, with nanoscale sensitivity, preneoplastic alterations in nuclear architecture of normal appearing cells undergoing malignant transformation both in animal models of carcinogenesis and human patients. In this talk I will present its theory and implementation from first principles and demonstrate its utility in a range of cancer settings. I will further present our recent results on testing the hypothesis that the ability of nanoNAM to capture early-stage malignant transformation is due to its sensitivity to the underlying aberrant structural alterations associated with chromatin remodeling during early stages of carcinogenesis. I will conclude by discussing its potential for use in the clinic.

We demonstrate a deep learning-based framework, called Fourier Imager Network (FIN), which achieves unparalleled generalization in end-to-end phase-recovery and hologram reconstruction. We used Fourier transform modules in FIN architecture, which process the spatial frequencies of the input images in a global receptive field and bring strong regularization and robustness to the hologram reconstruction task. We validated FIN by training it on human lung tissue samples and blindly testing it on human prostate, salivary gland, and Pap smear samples. FIN exhibits superior internal and external generalization compared with existing hologram reconstruction models, also achieving a ~50-fold acceleration in image inference speed.

The dielectric tensor is a physical quantity which characterizes birefringent materials with principal refractive indices and orientations of optic axes. Recently, three-dimensional dielectric tensor distribution was directly measured using dielectric tensor tomography (DTT). However, since the original DTT uses two cameras to acquire polarization-sensitive fields, position disagreement between the two fields deteriorates reconstruction quality. Here, we present multiplexed DTT using only one camera. To avoid the position disagreement, we exploit holographic multiplexing, interfering two orthogonally polarized reference beams with a sample beam on the camera. We validate the present method via measurement of anisotropic structures in liquid crystal particles.

We present the interferometric phase and absorption microscope (iPAM) platform for studying the biomechanics and fitness of sickle red blood cells (sRBCs). iPAM combines diffraction phase microscopy, tailored imaging media with customized optical properties, and extinction-based models to extract sRBC hemoglobin concentration and 3D structure from hundreds of cells in a single snapshot. Coupled with a customized microfluidic platform controlling cell deformation and oxygenation, we show that iPAM can provide a fitness index evaluating sRBC health that could act as an indicator detecting vaso-occlusive crisis risk and determine sickle cell disease treatment effectiveness.

Measuring cancer cell response to therapy is an essential component of developing new therapies or determining the optimal treatment for a particular patient. In this talk I will discuss the use of quantitative phase imaging (QPI) to measure the distribution of mass within single cells, and how this distribution changes over time. Based on this approach, my lab has developed new approaches to quantify the time-dependence and intracellular spatial distribution of cell response to small-molecule inhibitors. These data can be used to study basic cell physiology, as well as how cancer cells respond to therapies.

Accurate and rapid evaluation of dynamic immune status is critical to determine therapeutic modalities for sepsis patients, which is impeded by the limitations of conventional diagnostic tools. Here, we employ refractive index tomography to quantitatively assess the immune status of human monocytes in a label-free manner. Measurement of refractive index tomograms enabled quantifications of three-dimensional morphological parameters, which revealed a clear increment in lipid droplets content and intracellular inhomogeneities as the septic stage progresses. We leveraged these observations to engineer a deep-learning-based algorithm that predicts the immune status of monocytes, showing over 99 % blind test accuracy.

Cross-grating phase microscopy (CGM) is a wavefront imaging technique that can map the intensity and phase of a light beam with high sensitivity and spatial resolution. CGM is based on the association of a 2D-grating with a camera, separated by a millimetric distance. In this contribution, we will show how CGM can measure the dry mass of single neurites of neural cells in culture, and how it can highlight the subtle dynamics and mass transport within neurites, in a label-free manner.

We developed a large-scale single-cell intrinsic morphological profiling strategy using ultrahigh-throughput quantitative phase imaging (QPI) combined with a novel spinning on-the-fly cell-based assay platform. This integrated system demonstrates the unprecedented functional assay capability of QPI in not only scaling up the assay throughput, but also empowering new cytometric power to perform multiplexed live-cell drug screening (96 conditions in a single run) and genetic perturbation assay (by CRISPR). This platform thus allows generation of large cellular QPI datasets (4.85 TByte in this study) that could spearhead cost-effective label-free solutions for identifying disease-or gene-related cellular morphological phenotypes in therapeutics screening.

This conference presentation was prepared for the Quantitative Phase Imaging IX conference at SPIE BiOS, 2023.

This conference presentation was prepared for the Quantitative Phase Imaging IX conference at SPIE BiOS, 2023.

This conference presentation was prepared for the Quantitative Phase Imaging IX conference at SPIE BiOS, 2023.

This conference presentation was prepared for the Quantitative Phase Imaging IX conference at SPIE BiOS, 2023.