Search

Significance: Subretinal injection is an effective way of delivering transplant genes and cells to treat many degenerative retinal diseases. However, the technique requires high-dexterity and microscale precision of experienced surgeons, who have to overcome the physiological hand tremor and limited visualization of the subretinal space. Aim: To automatically guide the axial motion of microsurgical tools (i.e., a subretinal injector) with microscale precision in real time using a fiber-optic common-path swept-source optical coherence tomography distal sensor. Approach: We propose, implement, and study real-time retinal boundary tracking of A-scan optical coherence tomography (OCT) images using a convolutional neural network (CNN) for automatic depth targeting of a selected retinal boundary for accurate subretinal injection guidance. A simplified 1D U-net is used for the retinal layer segmentation on A-scan OCT images. A Kalman filter, combining retinal boundary position measurement by CNN and velocity measurement by cross correlation between consecutive A-scan images, is applied to optimally estimate the retinal boundary position. Unwanted axial motions of the surgical tools are compensated by a piezoelectric linear motor based on the retinal boundary tracking. Results: CNN-based segmentation on A-scan OCT images achieves the mean unsigned error (MUE) of ∼3 pixels (8.1 μm) using an ex vivo bovine retina model. GPU parallel computing allows real-time inference (∼2 ms) and thus real-time retinal boundary tracking. Involuntary tremors, which include low-frequency draft in hundreds of micrometers and physiological tremors in tens of micrometers, are compensated effectively. The standard deviations of photoreceptor (PR) and choroid (CH) boundary positions get as low as 10.8 μm when the depth targeting is activated. Conclusions: A CNN-based common-path OCT distal sensor successfully tracks retinal boundaries, especially the PR/CH boundary for subretinal injection, and automatically guides the tooltip’s axial position in real time. The microscale depth targeting accuracy of our system shows its promising possibility for clinical application.

Significance: Lupus nephritis (LuN) is a chronic inflammatory kidney disease. The cellular mechanisms by which LuN progresses to kidney failure are poorly characterized. Automated instance segmentation of immune cells in immunofluorescence images of LuN can probe these cellular interactions. Aim: Our specific goal is to quantify how sample fixation and staining panel design impact automated instance segmentation and characterization of immune cells. Approach: Convolutional neural networks (CNNs) were trained to segment immune cells in fluorescence confocal images of LuN biopsies. Three datasets were used to probe the effects of fixation methods on cell features and the effects of one-marker versus two-marker per cell staining panels on CNN performance. Results: Networks trained for multi-class instance segmentation on fresh-frozen and formalin-fixed, paraffin-embedded (FFPE) samples stained with a two-marker panel had sensitivities of 0.87 and 0.91 and specificities of 0.82 and 0.88, respectively. Training on samples with a one-marker panel reduced sensitivity (0.72). Cell size and intercellular distances were significantly smaller in FFPE samples compared to fresh frozen (Kolmogorov–Smirnov, p≪0.0001). Conclusions: Fixation method significantly reduces cell size and intercellular distances in LuN biopsies. The use of two markers to identify cell subsets showed improved CNN sensitivity relative to using a single marker.

Purpose: Image registration is an important aspect of medical image analysis and a key component in many analysis concepts. Applications include fusion of multimodal images, multi-atlas segmentation, and whole-body analysis. Deformable image registration is often computationally expensive, and the need for efficient registration methods is highlighted by the emergence of large-scale image databases, e.g., the UK Biobank, providing imaging from 100,000 participants. Approach: We present a heterogeneous computing approach, utilizing both the CPU and the graphics processing unit (GPU), to accelerate a previously proposed image registration method. The parallelizable task of computing the matching criterion is offloaded to the GPU, where it can be computed efficiently, while the more complex optimization task is performed on the CPU. To lessen the impact of data synchronization between the CPU and GPU, we propose a pipeline model, effectively overlapping computational tasks with data synchronization. The performance is evaluated on a brain labeling task and compared with a CPU implementation of the same method and the popular advanced normalization tools (ANTs) software. Results: The proposed method presents a speed-up by factors of 4 and 8 against the CPU implementation and the ANTs software, respectively. A significant improvement in labeling quality was also observed, with measured mean Dice overlaps of 0.712 and 0.701 for our method and ANTs, respectively. Conclusions: We showed that the proposed method compares favorably to the ANTs software yielding both a significant speed-up and an improvement in labeling quality. The registration method together with the proposed parallelization strategy is implemented as an open-source software package, deform.

Photoacoustic tomography (PAT) is an imaging technique that enables us to visualize vasculature networks at high resolution at depths much greater than optical microscopy.1,2 PAT can also resolve tissue chromophores such as oxyhemoglobin and deoxyhemoglobin, which allows estimation of tissue oxygentation levels.1,3 Accurate estimation of tissue oxygentation is very important in the fields of neuroimaging, cancer diagnostics, and critical care medicine.4,5 PAT involves collection of broadband acoustic waves generated by thermoelastic expansion of tissue upon absorption of light (illuminated at near-infrared wavelengths).6,7 PAT image reconstruction involves recovery of the acoustic source distribution from the detected broadband acoustic waves, which can be performed using analytical or model-based methods.8 Analytical methods are computationally less expensive compared to model-based image reconstruction.8,9 In contrary, model-based schemes are known to generate accurate images compared to analytical schemes, specifically in scenarios where irregular acquisition geometry is involved or limited data is acquired.9,10 Even with the model-based schemes, PAT inversion becomes difficult when the number of ultrasound detector positions is limited, and the situation gets aggravated due to the presence of noise in the data. More recently deep learning-based image postprocessing schemes have been developed for artifact removal and for limited data situations in PAT/microscopy;11–13 however, these schemes were trained on simulated data and generating large experimental data is very difficult. Segmentation approaches were proposed to delineate skin lining in mesoscopic photoacoustic images.14 Active contours-based image segmentation was used to obtain the PAT region of interest (ROI), the obtained ROI was then used to perform fluence correction to make PAT more quantitative.15,16 Different computational techniques have been reported in the literature to obtain vascular network in photoacoustic imaging. Vesselness filter has been proposed to enable accurate recovery of vasculature in PAT;17–19 however, questions have been raised on the accuracy of using the vesselness filter (due to generation of artificial vasculature).20 Probabilistic approach involving different steps such as smoothing and filtering the data, clustering, vessel-segmentation using clusters, and morphological filling was developed for segmenting vessels in photoacoustic images.21 More recently convolution neural network-based approaches were also developed for jointly performing the image reconstruction and segmentation.22 Further, deep learning-based approach was proposed for segmenting animal boundary using multi-modal photoacoustic and ultrasound data.23 Further, deep learning-based method has also been developed for animal brain imaging.24 Notably segmenting vasculature is very important in the context of diagnosis and surgical guidance, specifically for cardiovascular diseases.25,26 In this work, we developed a binary tomography approach wherein we assume the reconstruction distribution contains only two unknowns namely the background and the absorbers. The major contributions of this work include: (a) application of binary tomographic methods for photoacoustic tomographic imaging involving spherical wave propagation as opposed to straight line approximation applied in x-ray computed tomography; (b) showing the utility of binary tomographic methods for improving photoacoustic tomographic imaging of vasculature; and (c) establishing the binary tomographic method superiority in photoacoustic tomographic imaging for providing approximate priors to improve model-based image reconstruction. Binary tomography problem is solved based on a dual optimization approach with an auxiliary variable posed as a constrained optimization problem. We compared the binary tomography algorithm against standard backprojection approach and model-based reconstruction schemes such as Tikhonov regularization and sparsity-based regularization.27 The performance of these different algorithms was validated with in-silico data, physical phantom, and in-vivo brain vasculature imaging data.

Significance: Label-free quantitative phase imaging is a promising technique for the automatic detection of abnormal red blood cells (RBCs) in real time. Although deep-learning techniques can accurately detect abnormal RBCs from quantitative phase images efficiently, their applications in diagnostic testing are limited by the lack of transparency. More interpretable results such as morphological and biochemical characteristics of individual RBCs are highly desirable. Aim: An end-to-end deep-learning model was developed to efficiently discriminate thalassemic RBCs (tRBCs) from healthy RBCs (hRBCs) in quantitative phase images and segment RBCs for single-cell characterization. Approach: Two-dimensional quantitative phase images of hRBCs and tRBCs were acquired using digital holographic microscopy. A mask region-based convolutional neural network (Mask R-CNN) model was trained to discriminate tRBCs and segment individual RBCs. Characterization of tRBCs was achieved utilizing SHapley Additive exPlanation analysis and canonical correlation analysis on automatically segmented RBC phase images. Results: The implemented model achieved 97.8% accuracy in detecting tRBCs. Phase-shift statistics showed the highest influence on the correct classification of tRBCs. Associations between the phase-shift features and three-dimensional morphological features were revealed. Conclusions: The implemented Mask R-CNN model accurately identified tRBCs and segmented RBCs to provide single-RBC characterization, which has the potential to aid clinical decision-making.

Purpose: Hippocampus contouring for radiotherapy planning is performed on MR image data due to poor anatomical visibility on computed tomography (CT) data. Deep learning methods for direct CT hippocampus auto-segmentation exist, but use MR-based training contours. We investigate if these can be replaced by CT-based contours without loss in segmentation performance. This would remove the MR not only from inference but also from training. Approach: The hippocampus was contoured by medical experts on MR and CT data of 45 patients. Convolutional neural networks (CNNs) for hippocampus segmentation on CT were trained on CT-based or propagated MR-based contours. In both cases, their predictions were evaluated against the MR-based contours considered as the ground truth. Performance was measured using several metrics, including Dice score, surface distances, and contour Dice score. Bayesian dropout was used to estimate model uncertainty. Results: CNNs trained on propagated MR contours (median Dice 0.67) significantly outperform those trained on CT contours (0.59) and also experts contouring manually on CT (0.59). Differences between the latter two are not significant. Training on MR contours results in lower model uncertainty than training on CT contours. All contouring methods (manual or CNN) on CT perform significantly worse than a CNN segmenting the hippocampus directly on MR (median Dice 0.76). Additional data augmentation by rigid transformations improves the quantitative results but the difference remains significant. Conclusions: CT-based training contours for CT hippocampus segmentation cannot replace propagated MR-based contours without significant loss in performance. However, if MR-based contours are used, the resulting segmentations outperform experts in contouring the hippocampus on CT.

Significance: Detection and characterization of coronary atherosclerotic plaques often need reviews of a large number of optical coherence tomography (OCT) imaging slices to make a clinical decision. However, it is a challenge to manually review all the slices and consider the interrelationship between adjacent slices. Approach: Inspired by the recent success of deep convolutional network on the classification of medical images, we proposed a ResNet-3D network for classification of coronary plaque calcification in OCT pullbacks. The ResNet-3D network was initialized with a trained ResNet-50 network and a three-dimensional convolution filter filled with zeros padding and non-zeros padding with a convolutional filter. To retrain ResNet-50, we used a dataset of ∼4860 OCT images, derived by 18 entire pullbacks from different patients. In addition, we investigated a two-phase training method to address the data imbalance. For an improved performance, we evaluated different input sizes for the ResNet-3D network, such as 3, 5, and 7 OCT slices. Furthermore, we integrated all ResNet-3D results by majority voting. Results: A comparative analysis proved the effectiveness of the proposed ResNet-3D networks against ResNet-2D network in the OCT dataset. The classification performance (F1-scores=94% for non-zeros padding and F1-score=96% for zeros padding) demonstrated the potential of convolutional neural networks (CNNs) in classifying plaque calcification. Conclusions: This work may provide a foundation for further work in extending the CNN to voxel segmentation, which may lead to a supportive diagnostic tool for assessment of coronary plaque vulnerability.

New increasingly complex in vitro cancer cell models are being developed. These new models seem to represent the cell behavior in vivo more accurately and have better physiological relevance than prior models. An efficient testing method for selecting the most optimal drug treatment does not exist to date. One proposed solution to the problem involves isolation of cancer cells from the patients' cancer tissue, after which they are exposed to potential drugs alone or in combinations to find the most optimal medication. To achieve this goal, methods that can efficiently quantify and analyze changes in tested cell are needed. Our study aimed to detect and segment cells and structures from cancer cell cultures grown on vascular structures in phase-contrast microscope images using U-Net neural networks to enable future drug efficacy assessments. We cultivated prostate carcinoma cell lines PC3 and LNCaP on the top of a matrix containing vascular structures. The cells were imaged with a Cell-IQ phase-contrast microscope. Automatic analysis of microscope images could assess the efficacy of tested drugs. The dataset included 36 RGB images and ground-truth segmentations with mutually not exclusive classes. The used method could distinguish vascular structures, cells, spheroids, and cell matter around spheroids in the test images. Some invasive spikes were also detected, but the method could not distinguish the invasive cells in the test images. publishedVersion

Significance: Machine learning is increasingly being applied to the classification of microscopic data. In order to detect some complex and dynamic cellular processes, time-resolved live-cell imaging might be necessary. Incorporating the temporal information into the classification process may allow for a better and more specific classification. Aim: We propose a methodology for cell classification based on the time-lapse quantitative phase images (QPIs) gained by digital holographic microscopy (DHM) with the goal of increasing performance of classification of dynamic cellular processes. Approach: The methodology was demonstrated by studying epithelial-mesenchymal transition (EMT) which entails major and distinct time-dependent morphological changes. The time-lapse QPIs of EMT were obtained over a 48-h period and specific novel features representing the dynamic cell behavior were extracted. The two distinct end-state phenotypes were classified by several supervised machine learning algorithms and the results were compared with the classification performed on single-time-point images. Results: In comparison to the single-time-point approach, our data suggest the incorporation of temporal information into the classification of cell phenotypes during EMT improves performance by nearly 9% in terms of accuracy, and further indicate the potential of DHM to monitor cellular morphological changes. Conclusions: Proposed approach based on the time-lapse images gained by DHM could improve the monitoring of live cell behavior in an automated fashion and could be further developed into a tool for high-throughput automated analysis of unique cell behavior. (C) The Authors. Published by SPIE under a Creative Commons Attribution 4.0 Unported License.

Fluorescence microscopy plays a vital role in understanding the subcellular structures of living cells. However, it requires considerable effort in sample preparation related to chemical fixation, staining, cost, and time. To reduce those factors, we present a virtual fluorescence staining method based on deep neural networks (VirFluoNet) to transform fluorescence images of molecular labels into other molecular fluorescence labels in the same field-of-view. To achieve this goal, we develop and train a conditional generative adversarial network (cGAN) to perform digital fluorescence imaging demonstrated on human osteosarcoma U2OS cell fluorescence images captured under Cell Painting staining protocol. A detailed comparative analysis is also conducted on the performance of the cGAN network between predicting fluorescence channels based on phase contrast or based on another fluorescence channel using human breast cancer MDA-MB-231 cell line as a test case. In addition, we implement a deep learning model to perform autofocusing on another human U2OS fluorescence dataset as a preprocessing step to defocus an out-focus channel in U2OS dataset. A quantitative index of image prediction error is introduced based on signal pixel-wise spatial and intensity differences with ground truth to evaluate the performance of prediction to high-complex and throughput fluorescence. This index provides a rational way to perform image segmentation on error signals and to understand the likelihood of mis-interpreting biology from the predicted image. In total, these findings contribute to the utility of deep learning image regression for fluorescence microscopy datasets of biological cells, balanced against savings of cost, time, and experimental effort. Furthermore, the approach introduced here holds promise for modeling the internal relationships between organelles and biomolecules within living cells., 20 pages, 9 figures