arXiv subjects

Natan T. Shaked

Publications and source records attributed to Natan T. Shaked.

At least 19 recordsLinked to original sources

arXiv2026

PRISM: Distribution-Gated Flow Matching for Controllable Unpaired Image Translation

Unpaired image-to-image translation must decide, per image, what to change and what to preserve without paired supervision. Many diffusion-based unpaired translators control preservation through a single global noise or guidance value applied across the image, which cannot separate content to keep from appearance to change. We present PRISM, a GAN-free flow-matching framework that replaces this global control with a learned per-feature gate. The gate's spatial prior is derived from each source feature's standardized distance to the target feature distribution, so features far from the target are freed while target-consistent features are preserved. The same gate controls both the initialization, which mixes the real source latent with a task-matched corruption, and the transport timing during Ordinary Differential Equation (ODE) integration. The corruption is matched to the task, content-anchored (AdaIN) for structure-preserving translation and partially anchored for structure-changing translation, and the gate can be overridden locally at inference from text or a detector without retraining, preserving important structures of the original image while still generating realistic results. We evaluate PRISM on five natural and biomedical benchmarks (AFHQ cat->dog, CelebA-HQ appearance translation, day->night relighting, virtual staining, and breast frozen->permanent histopathology). Among the evaluated methods under a shared same-split protocol, PRISM attains the best Inception FID and KID on four benchmarks and a competitive result on the fifth, and on histopathology yields the nuclei-count ratio closest to the ideal, supporting a favorable balance between target realism and structural preservation.

cs.CV↗

arXiv2026

CAFlow: Adaptive-Depth Single-Step Flow Matching for Efficient Histopathology Super-Resolution

In digital pathology, whole-slide images routinely exceed gigapixel resolution, making computationally intensive generative super-resolution (SR) impractical for routine deployment. We introduce CAFlow, an adaptive-depth single-step flow-matching framework that routes each image tile to the shallowest network exit that preserves reconstruction quality. CAFlow performs flow matching in pixel-unshuffled rearranged space, reducing spatial computation by 16x while enabling direct inference. We show that dedicating half of training to exact t=0 samples is essential for single-step quality (-1.5 dB without it). The backbone, FlowResNet (1.90M parameters), mixes convolution and window self-attention blocks across four early exits spanning 3.1 to 13.3 GFLOPs. A lightweight exit classifier (~6K parameters) achieves 33% compute savings at only 0.12 dB cost. On multi-organ histopathology x4 SR, adaptive routing achieves 31.72 dB PSNR versus 31.84 dB at full depth, while the shallowest exit exceeds bicubic by +1.9 dB at 2.8x less compute than SwinIR-light. The method generalizes to held-out colon tissue with minimal quality loss (-0.02 dB), and at x8 upscaling it outperforms all comparable-compute baselines while remaining competitive with the much larger SwinIR-Medium model. Downstream nuclei segmentation confirms preservation of clinically relevant structure. The model trains in under 5 hours on a single GPU, and adaptive routing can reduce whole-slide inference from minutes to seconds.

cs.CV↗

arXiv2025

Hierarchical Sparse Attention Framework for Computationally Efficient Classification of Biological Cells

We present SparseAttnNet, a new hierarchical attention-driven framework for efficient image classification that adaptively selects and processes only the most informative pixels from images. Traditional convolutional neural networks typically process the entire images regardless of information density, leading to computational inefficiency and potential focus on irrelevant features. Our approach leverages a dynamic selection mechanism that uses coarse attention distilled by fine multi-head attention from the downstream layers of the model, allowing the model to identify and extract the most salient k pixels, where k is adaptively learned during training based on loss convergence trends. Once the top-k pixels are selected, the model processes only these pixels, embedding them as words in a language model to capture their semantics, followed by multi-head attention to incorporate global context. For biological cell images, we demonstrate that SparseAttnNet can process approximately 15% of the pixels instead of the full image. Applied to cell classification tasks using white blood cells images from the following modalities: optical path difference (OPD) images from digital holography for stain-free cells, images from motion-sensitive (event) camera from stain-free cells, and brightfield microscopy images of stained cells, For all three imaging modalities, SparseAttnNet achieves competitive accuracy while drastically reducing computational requirements in terms of both parameters and floating-point operations per second, compared to traditional CNNs and Vision Transformers. Since the model focuses on biologically relevant regions, it also offers improved explainability. The adaptive and lightweight nature of SparseAttnNet makes it ideal for deployment in resource-constrained and high-throughput settings, including imaging flow cytometry.

eess.IV↗

arXiv2024

Enhancing frozen histological section images using permanent-section-guided deep learning with nuclei attention

In histological pathology, frozen sections are often used for rapid diagnosis during surgeries, as they can be produced within minutes. However, they suffer from artifacts and often lack crucial diagnostic details, particularly within the cell nuclei region. Permanent sections, on the other hand, contain more diagnostic detail but require a time-intensive preparation process. Here, we present a generative deep learning approach to enhance frozen section images by leveraging guidance from permanent sections. Our method places a strong emphasis on the nuclei region, which contains critical information in both frozen and permanent sections. Importantly, our approach avoids generating artificial data in blank regions, ensuring that the network only enhances existing features without introducing potentially unreliable information. We achieve this through a segmented attention network, incorporating nuclei-segmented images during training and adding an additional loss function to refine the nuclei details in the generated permanent images. We validated our method across various tissues, including kidney, breast, and colon. This approach significantly improves histological efficiency and diagnostic accuracy, enhancing frozen section images within seconds, and seamlessly integrating into existing laboratory workflows.

eess.IV↗

arXiv2023

Super resolution of histopathological frozen sections via deep learning preserving tissue structure

Histopathology plays a pivotal role in medical diagnostics. In contrast to preparing permanent sections for histopathology, a time-consuming process, preparing frozen sections is significantly faster and can be performed during surgery, where the sample scanning time should be optimized. Super-resolution techniques allow imaging the sample in lower magnification and sparing scanning time. In this paper, we present a new approach to super resolution for histopathological frozen sections, with focus on achieving better distortion measures, rather than pursuing photorealistic images that may compromise critical diagnostic information. Our deep-learning architecture focuses on learning the error between interpolated images and real images, thereby it generates high-resolution images while preserving critical image details, reducing the risk of diagnostic misinterpretation. This is done by leveraging the loss functions in the frequency domain, assigning higher weights to the reconstruction of complex, high-frequency components. In comparison to existing methods, we obtained significant improvements in terms of Structural Similarity Index (SSIM) and Peak Signal-to-Noise Ratio (PSNR), as well as indicated details that lost in the low-resolution frozen-section images, affecting the pathologist's clinical decisions. Our approach has a great potential in providing more-rapid frozen-section imaging, with less scanning, while preserving the high resolution in the imaged sample.

eess.IV↗

arXiv2021

Real-time stain-free classification of cancer cells and blood cells using interferometric phase microscopy and machine learning

We present a method for a real time visualization and automatic processing for detection and classification of untouched cancer cells in blood during stain free imaging flow cytometry using digital holographic microscopy and machine learning in throughput of 15 cells per second. As a preliminary model for circulating tumor cells in blood, we automatically classified primary and metastatic colon cancer cells, where the two types of cancer cells were isolated from the same individual, as well as four types of blood cells. We used low-coherence off-axis interferometric phase microscopy and a microfluidic channel to quantitatively image cells during flow. The acquired images were processed and classified based on their morphology and quantitative phase features during the cell flow. We achieved high accuracy of 92.56 percent for distinguishing between the cells, paving the way for future automatic label-free cancer cell classification in blood samples.

physics.bio-ph↗

arXiv2020

High-resolution 4D acquisition of freely swimming human sperm cells without staining

We present a new acquisition method that enables high-resolution, fine-detail full reconstruction of the three-dimensional movement and structure of individual human sperm cells swimming freely. We achieve both retrieval of the three-dimensional refractive-index profile of the sperm head, revealing its fine internal organelles and time-varying orientation, and the detailed four-dimensional localization of the thin, highly-dynamic flagellum of the sperm cell. Live human sperm cells were acquired during free swim using a high-speed off-axis holographic system that does not require any moving elements or cell staining. The reconstruction is based solely on the natural movement of the sperm cell and a novel set of algorithms, enabling the detailed four-dimensional recovery. Using this refractive-index imaging approach, we believe we have detected an area in the cell that is attributed to the centriole. This method has great potential for both biological assays and clinical use of intact sperm cells.

physics.bio-ph↗

arXiv2019

HoloStain: Holographic virtual staining of individual biological cells

Many medical and biological protocols for analyzing individual biological cells involve morphological evaluation based on cell staining, designed to enhance imaging contrast and enable clinicians and biologists to differentiate between various cell organelles. However, cell staining is not always allowed in certain medical procedures. In other cases, staining may be time consuming or expensive to implement. Here, we present a new deep-learning approach, called HoloStain, which converts images of isolated biological cells acquired without staining by holographic microscopy to their virtually stained images. We demonstrate this approach for human sperm cells, as there is a well-established protocol and global standardization for characterizing the morphology of stained human sperm cells for fertility evaluation, but, on the other hand, staining might be cytotoxic and thus is not allowed during human in vitro fertilization (IVF). We use deep convolutional Generative Adversarial Networks (DCGANs) with training that is based on both the quantitative phase images and two gradient phase images, all extracted from the digital holograms of the stain-free cells, with the ground truth of bright-field images of the same cells that subsequently underwent chemical staining. To validate the quality of our virtual staining approach, an experienced embryologist analyzed the unstained cells, the virtually stained cells, and the chemically stained sperm cells several times in a blinded and randomized manner. We obtained a 5-fold recall (sensitivity) improvement in the analysis results. With the introduction of simple holographic imaging methods in clinical settings, the proposed method has a great potential to become a common practice in human IVF procedures, as well as to significantly simplify and facilitate other cell analyses and techniques such as imaging flow cytometry.

physics.bio-ph↗

arXiv2019

Integral refractive index imaging of flowing cell nuclei using quantitative phase microscopy combined with fluorescence microscopy

We suggest a new multimodal imaging technique for quantitatively measuring the integral (thickness-average) refractive index of the nuclei of live biological cells in suspension. For this aim, we combined quantitative phase microscopy with simultaneous 2-D fluorescence microscopy. We used 2-D fluorescence microscopy to localize the nucleus inside the quantitative phase map of the cell, as well as for measuring the nucleus radii. As verified offline by both 3-D confocal fluorescence microscopy and by 2-D fluorescence microscopy while rotating the cells during flow, the nucleus of cells in suspension that are not during division can be assumed to be an ellipsoid. The entire shape of a cell in suspension can be assumed to be a sphere. Then, the cell and nucleus 3-D shapes can be evaluated based on their in-plain radii available from the 2-D phase and fluorescent measurements, respectively. Finally, the nucleus integral refractive index profile is calculated. We demonstrate the new technique on cancer cells, obtaining nucleus refractive index values that are lower than those of the cytoplasm, coinciding with recent findings. We believe that the proposed technique has the potential to be used for flow cytometry, where full 3-D refractive index tomography is too slow to be implemented during flow.

physics.bio-ph↗

arXiv2019

Four dimensional phase unwrapping of dynamic objects in digital holography

We present a new four-dimensional phase unwrapping approach for time-lapse quantitative phase microscopy, which allows reconstruction of optically thick objects that are optically thin in a certain temporal point and angular view. We thus use all four dimensions of the dynamic quantitative phase profile acquired, including the angular dimension and the temporal dimension, in addition to the x-y dimensions. We first demonstrate the capabilities of this algorithm on simulative data, enabling the quantification of the reconstruction quality relative to both the ground truth and existing unwrapping approaches. Then, we demonstrate the applicability of the proposed four-dimensional phase unwrapping algorithm by experimentally capturing a dual-angular dynamic off-axis hologram with simultaneous recording of two angular views, using multiplexing of two off-axis holograms into a single multiplexed hologram.

physics.optics↗

arXiv2019

Quantitative phase microscopy spatial signatures of cancer cells

We present cytometric classification of live healthy and cancer cells by using the spatial morphological and textural information found in the label-free quantitative phase images of the cells. We compare both healthy cells to primary tumor cell and primary tumor cells to metastatic cancer cells, where tumor biopsies and normal tissues were isolated from the same individuals. To mimic analysis of liquid biopsies by flow cytometry, the cells were imaged while unattached to the substrate. We used low-coherence off-axis interferometric phase microscopy setup, which allows a single-exposure acquisition mode, and thus is suitable for quantitative imaging of dynamic cells during flow. After acquisition, the optical path delay maps of the cells were extracted, and used to calculate 15 parameters derived from cellular 3-D morphology and texture. Upon analyzing tens of cells in each group, we found high statistical significance in the difference between the groups in most of the parameters calculated, with the same trends for all statistically significant parameters. Furthermore, a specially designed machine learning algorithm, implemented on the phase map extracted features, classified the correct cell type (healthy/cancer/metastatic) with 81%-93% sensitivity and 81%-99% specificity. The quantitative phase imaging approach for liquid biopsies presented in this paper could be the basis for advanced techniques of staging freshly isolated live cancer cells in imaging flow cytometers.

q-bio.QM↗

arXiv2019

Review on methods of solving the refractive index-thickness coupling problem in digital holographic microscopy of biological cells

Digital holographic microscopy is a thriving imaging modality that attracted considerable research interest in quantitative biological cell imaging due to its ability to not only create excellent label-free contrast, but also supply valuable physical information regarding the density and dimensions of the sample with nanometer-scale axial sensitivity. This technique records the interference pattern between a sample beam and a reference beam, and by digitally processing it, one can reconstruct the optical path delay between these beams. Per each spatial point, the optical path delay map is proportional to the product of the sample physical thickness and the integral refractive index of the sample. Since the refractive index of the cell indicates its contents without the need for labeling, it is highly beneficial to decouple cell physical thickness from its refractive index profile. This manuscript reviews various approaches of extracting the refractive index from digital holographic microscopy measurements of cells. As soon as the refractive index of the cell is available, it can be used for either biological assays or medical diagnosis, as reviewed in this manuscript.

physics.optics↗

arXiv2019

Simultanous two-wavelength phase unwrapping using external module for multiplexing off-axis holography

We present a dual-wavelength external holographic microscopy module for quantitative phase imaging of 3D structures with extended thickness range. This is done by simultaneous acquisition of two off-axis interferograms, each of which at a different wavelength, and generation of a synthetic wavelength, which is larger than the sample optical thickness, allowing two-wavelength unwrapping. The simultaneous acquisition is carried out by using optical multiplexing of the two interferograms onto the camera, where each of them has orthogonal off-axis interference fringe direction in relation to the other one. We used the system to quantitatively image a 7.96 um step target and 30.5 um circular copper pillars.

physics.optics↗

arXiv2019

Six-pack off-axis holography

We present a new holographic concept, named six-pack holography (6PH), in which we compress six off-axis holograms into a single multiplexed off-axis hologram without loss of magnification or resolution. The multiplexed hologram contains straight off-axis fringes with six different orientations, and can be generated optically or digitally. We show that since the six different complex wavefronts do not overlap in the spatial frequency domain, they can be fully reconstructed. 6PH allows more than 50% improvement in the spatial bandwidth consumption when compared to the best multiplexing method proposed so far. We expect the 6PH concept to be useful for a variety of applications, such as field-of-view multiplexing, wavelength multiplexing, temporal multiplexing, multiplexing for super-resolution imaging, and others.

physics.optics↗

arXiv2019

Simultaneous off-axis multiplexed holography and regular fluorescence microscopy of biological cells

We present a new technique for obtaining simultaneous multimodal quantitative phase and fluorescence microscopy of biological cells, providing both quantitative phase imaging and molecular specificity using a single camera. Our system is based on an interferometric multiplexing module, externally positioned at the exit of an optical microscope. In contrast to previous approaches, the presented technique allows conventional fluorescence imaging, rather than interferometric off-axis fluorescence imaging. We demonstrate the presented technique for imaging fluorescent beads and live biological cells.

physics.optics↗

arXiv2019

Is multiplexed off-axis holography for quantitative phase imaging more spatial bandwidth-efficient than on-axis holography?

Digital holographic microcopy is a thriving imaging modality that attracts considerable research interest due to its ability to not only create excellent label-free contrast, but also supply valuable physical information regarding the density and dimensions of the sample with nanometer-scale axial sensitivity. Three basic holographic recording geometries currently exist, including on-axis, off-axis and slightly off-axis holography, each of them enabling a variety of architectures in terms of bandwidth use and compression capacity. Specifically, off-axis holography and slightly off-axis holography allow spatial hologram multiplexing, enabling compressing more information into the same digital hologram. In this paper, we define an efficiency score used to analyze the various possible architectures, and compare the signal-to-noise ratio and mean squared error obtained using each of them, determining the optimal holographic method.

physics.optics↗

arXiv2019

Simultaneous three-wavelength unwrapping using external digital holographic multiplexing module

We present an external interferometric setup that is able to simultaneously acquire three wavelengths of the same sample instance without scanning or multiple exposures. This setup projects onto the monochrome digital camera three off-axis holograms with rotated fringe orientations, each from a different wavelength channel, without overlap in the spatial-frequency domain, and thus allows the full reconstruction of the three complex wavefronts from the three wavelength channels. We use this new setup for three-wavelength phase unwrapping, allowing phase imaging of thicker objects than possible with a single wavelength, but without the increased level of noise. We demonstrate the proposed technique for micro-channel profiling and label-free cell imaging.

physics.optics↗

arXiv2019

Single-exposure full-field multi-depth imaging using low-coherence holographic multiplexing

We present a new interferometric imaging approach that allows for multiple-depth imaging in a single acquisition, using off-axis low-coherence holographic multiplexing. This technique enables sectioned imaging of multiple slices within a thick sample, in a single image acquisition. Each slice has a distinct off-axis interference fringe orientation indicative of its axial location, and the camera acquires the multiplexed hologram containing the different slices at once. We demonstrate the proposed technique for amplitude and phase imaging of optically thick samples.

physics.optics↗