Showing papers in "Biomedical Optics Express in 2019"
TL;DR: The experimental results indicate that the CNN based system can be a valuable option for the design of a computer aid system for the detection of glaucoma in large-scale screening programs.
Abstract: Glaucoma detection in color fundus images is a challenging task that requires expertise and years of practice. In this study we exploited the application of different Convolutional Neural Networks (CNN) schemes to show the influence in the performance of relevant factors like the data set size, the architecture and the use of transfer learning vs newly defined architectures. We also compared the performance of the CNN based system with respect to human evaluators and explored the influence of the integration of images and data collected from the clinical history of the patients. We accomplished the best performance using a transfer learning scheme with VGG19 achieving an AUC of 0.94 with sensitivity and specificity ratios similar to the expert evaluators of the study. The experimental results using three different data sets with 2313 images indicate that this solution can be a valuable option for the design of a computer aid system for the detection of glaucoma in large-scale screening programs.
131 citations
TL;DR: This review provides an overview of the current state-of-the-art methods used for both ex- vivo and in-vivo optical clearing of biological tissues, classified based on the tissue size and type for each specific application.
Abstract: Every optical imaging technique is limited in its penetration depth by scattering occurring in biological tissues. Possible solutions to overcome this problem consist of limiting the detrimental effects of scattering by reducing optical inhomogeneities within the sample. This can be achieved either by using physical methods (such as refractive index matching solutions) or by chemical methods (such as the removal of scatterers), based on tissue transformation protocols. This review provides an overview of the current state-of-the-art methods used for both ex-vivo and in-vivo optical clearing of biological tissues. We start with a brief history of the development of the most widespread clearing methods across the new millennium, then we describe the working principles of both physical and chemical methods. Clearing methods are then reviewed, pointing the attention of the reader on both physical and chemical methods, classified based on the tissue size and type for each specific application. A small section is reserved for methods that have already found in-vivo applications at the research level. Finally, a detailed discussion highlighting both the most relevant results achieved and the new ongoing developments in this field is reported in the last part, together with future perspectives for the clearing methodology.
121 citations
TL;DR: The results show that deep learning algorithms can be used for OCT image segmentation and could be applied in various clinical settings and in particular, CorneaNet could be use for early detection of keratoconus and more generally to study other diseases altering corneal morphology.
Abstract: Deep learning has dramatically improved object recognition, speech recognition, medical image analysis and many other fields. Optical coherence tomography (OCT) has become a standard of care imaging modality for ophthalmology. We asked whether deep learning could be used to segment cornea OCT images. Using a custom-built ultrahigh-resolution OCT system, we scanned 72 healthy eyes and 70 keratoconic eyes. In total, 20,160 images were labeled and used for the training in a supervised learning approach. A custom neural network architecture called CorneaNet was designed and trained. Our results show that CorneaNet is able to segment both healthy and keratoconus images with high accuracy (validation accuracy: 99.56%). Thickness maps of the three main corneal layers (epithelium, Bowman's layer and stroma) were generated both in healthy subjects and subjects suffering from keratoconus. CorneaNet is more than 50 times faster than our previous algorithm. Our results show that deep learning algorithms can be used for OCT image segmentation and could be applied in various clinical settings. In particular, CorneaNet could be used for early detection of keratoconus and more generally to study other diseases altering corneal morphology.
100 citations
TL;DR: This work combines a generative adversarial network (GAN) with light microscopy to achieve deep learning super-resolution under a large field of view (FOV) and proposes an image degrading model to generate low resolution images for training, making this approach free from the complex image registration during training data set preparation.
Abstract: We combine a generative adversarial network (GAN) with light microscopy to achieve deep learning super-resolution under a large field of view (FOV). By appropriately adopting prior microscopy data in an adversarial training, the neural network can recover a high-resolution, accurate image of new specimen from its single low-resolution measurement. Its capacity has been broadly demonstrated via imaging various types of samples, such as USAF resolution target, human pathological slides, fluorescence-labelled fibroblast cells, and deep tissues in transgenic mouse brain, by both wide-field and light-sheet microscopes. The gigapixel, multi-color reconstruction of these samples verifies a successful GAN-based single image super-resolution procedure. We also propose an image degrading model to generate low resolution images for training, making our approach free from the complex image registration during training data set preparation. After a well-trained network has been created, this deep learning-based imaging approach is capable of recovering a large FOV (~95 mm2) enhanced resolution of ~1.7 μm at high speed (within 1 second), while not necessarily introducing any changes to the setup of existing microscopes.
94 citations
TL;DR: The decreases in retinal vessel density, choroid capillary flow, and blood flow area are closely related to coronary artery and branch stenosis.
Abstract: To reveal the association between retinal microvasculature changes and coronary heart disease (CHD), we assessed the full retinal thicknesses of eight areas, the vessel density of four layers (consisting of nine areas) and the flow area in two layers with optical coherence tomography angiography (OCTA) in CHD patients and healthy controls. The mean vessel density of several layers was significantly lower in patients. The difference in choroid capillary flow (negative correlation) between the two groups was significant. Decreased vessel density and blood flow were associated with coronary artery and branch stenosis. The decreases in retinal vessel density, choroidal vessel density, and blood flow area are closely related to coronary artery and branch stenosis.
86 citations
TL;DR: A 4-class classification problem to automatically detect choroidal neovascularization, diabetic macular edema, DRUSEN, and NORMAL in optical coherence tomography (OCT) images and the effect of the integration of retinal OCT images and medical history data from patients on model performance is explored.
Abstract: Retinal disease classification is a significant problem in computer-aided diagnosis (CAD) for medical applications. This paper is focused on a 4-class classification problem to automatically detect choroidal neovascularization (CNV), diabetic macular edema (DME), DRUSEN, and NORMAL in optical coherence tomography (OCT) images. The proposed classification algorithm adopted an ensemble of four classification model instances to identify retinal OCT images, each of which was based on an improved residual neural network (ResNet50). The experiment followed a patient-level 10-fold cross-validation process, on development retinal OCT image dataset. The proposed approach achieved 0.973 (95% confidence interval [CI], 0.971–0.975) classification accuracy, 0.963 (95% CI, 0.960–0.966) sensitivity, and 0.985 (95% CI, 0.983–0.987) specificity at the B-scan level, achieving a matching or exceeding performance to that of ophthalmologists with significant clinical experience. Other performance measures used in the study were the area under receiver operating characteristic curve (AUC) and kappa value. The observations of the study implied that multi-ResNet50 ensembling was a useful technique when the availability of medical images was limited. In addition, we performed qualitative evaluation of model predictions, and occlusion testing to understand the decision-making process of our model. The paper provided an analytical discussion on misclassification and pathology regions identified by the occlusion testing also. Finally, we explored the effect of the integration of retinal OCT images and medical history data from patients on model performance.
82 citations
TL;DR: The review concludes by outlining exciting technological prospects of en face OCT based both on time as well as on Fourier domain OCT, which is in particular of advantage for OCM or OCT angiography.
Abstract: A review on the technological development of en face optical coherence tomography (OCT) and optical coherence microscopy (OCM) is provided The terminology originally referred to time domain OCT, where the preferential scanning was performed in the en face plane Potentially the fastest realization of en face image recording is full-field OCT, where the full en face plane is illuminated and recorded simultaneously The term has nowadays been adopted for high-speed Fourier domain approaches, where the en face image is reconstructed from full 3D volumes either by direct slicing or through axial projection in post processing The success of modern en face OCT lies in its immediate and easy image interpretation, which is in particular of advantage for OCM or OCT angiography Applications of en face OCT with a focus on ophthalmology are presented The review concludes by outlining exciting technological prospects of en face OCT based both on time as well as on Fourier domain OCT
77 citations
TL;DR: High-resolution LFM is reported for live-cell imaging with a resolution of 300-700 nm in all three dimensions, an imaging depth of several micrometers, and a volume acquisition time of milliseconds, demonstrating the technique by imaging various cellular dynamics and structures and tracking single particles.
Abstract: Visualizing diverse anatomical and functional traits that span many spatial scales with high spatio-temporal resolution provides insights into the fundamentals of living organisms. Light-field microscopy (LFM) has recently emerged as a scanning-free, scalable method that allows for high-speed, volumetric functional brain imaging. Given those promising applications at the tissue level, at its other extreme, this highly-scalable approach holds great potential for observing structures and dynamics in single-cell specimens. However, the challenge remains for current LFM to achieve a subcellular level, near-diffraction-limited 3D spatial resolution. Here, we report high-resolution LFM (HR-LFM) for live-cell imaging with a resolution of 300-700 nm in all three dimensions, an imaging depth of several micrometers, and a volume acquisition time of milliseconds. We demonstrate the technique by imaging various cellular dynamics and structures and tracking single particles. The method may advance LFM as a particularly useful tool for understanding biological systems at multiple spatio-temporal levels.
72 citations
TL;DR: A localized surface plasmon resonance -based biosensor is demonstrated that accurately detects and measures the concentration of cholesterol and the limit of detection is found to be 53.1 nM.
Abstract: Accurate cholesterol level measurement plays an important role in the diagnosis of severe diseases such as cardiovascular diseases, hypertension, anemia, myxedemia, hyperthyroidism, coronary artery illness. Traditionally, electrochemical sensors have been employed to detect the cholesterol level. However, these sensors have limitations in terms of sensitivity and selectivity. In this paper, a localized surface plasmon resonance (LSPR) -based biosensor is demonstrated that accurately detects and measures the concentration of cholesterol. In the present study, a tapered optical fiber-based sensor probe is developed using gold nanoparticles (AuNPs) and cholesterol oxidase (ChOx) to increase the sensitivity and selectivity of the sensor. Synthesized AuNPs were characterized by UV-visible spectrophotometer, transmission electron microscope (TEM), and energy dispersive X-ray spectroscopy (EDS). Further, coating of AuNPs over fiber was confirmed by scanning electron microscope (SEM). The developed sensor demonstrates for a clinically important cholesterol range of 0 to 10 mM, and the limit of detection is found to be 53.1 nM.
66 citations
TL;DR: A compact, hand-held hyperspectral imaging system for 2D neural network-based visualization of skin chromophores and blood oxygenation and enables a tool combining both the speed of an artificial neural network processing and the accuracy and flexibility of advanced Monte Carlo modeling.
Abstract: We developed a compact, hand-held hyperspectral imaging system for 2D neural network-based visualization of skin chromophores and blood oxygenation. State-of-the-art micro-optic multichannel matrix sensor combined with the tunable Fabry-Perot micro interferometer enables a portable diagnostic device sensitive to the changes of the oxygen saturation as well as the variations of blood volume fraction of human skin. Generalized object-oriented Monte Carlo model is used extensively for the training of an artificial neural network utilized for the hyperspectral image processing. In addition, the results are verified and validated via actual experiments with tissue phantoms and human skin in vivo. The proposed approach enables a tool combining both the speed of an artificial neural network processing and the accuracy and flexibility of advanced Monte Carlo modeling. Finally, the results of the feasibility studies and the experimental tests on biotissue phantoms and healthy volunteers are presented.
66 citations
TL;DR: This work presents the use of a deep learning algorithm to significantly improve the SNR of SRS images, based on a U-Net convolutional neural network and significantly outperforms existing denoising algorithms.
Abstract: Stimulated Raman scattering (SRS) microscopy is a label-free quantitative chemical imaging technique that has demonstrated great utility in biomedical imaging applications ranging from real-time stain-free histopathology to live animal imaging. However, similar to many other nonlinear optical imaging techniques, SRS images often suffer from low signal to noise ratio (SNR) due to absorption and scattering of light in tissue as well as the limitation in applicable power to minimize photodamage. We present the use of a deep learning algorithm to significantly improve the SNR of SRS images. Our algorithm is based on a U-Net convolutional neural network (CNN) and significantly outperforms existing denoising algorithms. More importantly, we demonstrate that the trained denoising algorithm is applicable to images acquired at different zoom, imaging power, imaging depth, and imaging geometries that are not included in the training. Our results identify deep learning as a powerful denoising tool for biomedical imaging at large, with potential towards in vivo applications, where imaging parameters are often variable and ground-truth images are not available to create a fully supervised learning training set.
TL;DR: This technique packages the quantitative, real-time sub-cellular imaging capabilities of QPI into a flexible configuration, opening the door for truly non-invasive, label-free, tomographic quantitative phase imaging of unaltered thick, scattering specimens.
Abstract: Quantitative phase imaging (QPI) is an important tool in biomedicine that allows for the microscopic investigation of live cells and other thin, transparent samples. Importantly, this technology yields access to the cellular and sub-cellular structure and activity at nanometer scales without labels or dyes. Despite this unparalleled ability, QPI’s restriction to relatively thin samples severely hinders its versatility and overall utility in biomedicine. Here we overcome this significant limitation of QPI to enable the same rich level of quantitative detail in thick scattering samples. We achieve this by first illuminating the sample in an epi-mode configuration and using multiple scattering within the sample—a hindrance to conventional transmission imaging used in QPI—as a source of transmissive illumination from within. Second, we quantify phase via deconvolution by modeling the transfer function of the system based on the ensemble average angular distribution of light illuminating the sample at the focal plane. This technique packages the quantitative, real-time sub-cellular imaging capabilities of QPI into a flexible configuration, opening the door for truly non-invasive, label-free, tomographic quantitative phase imaging of unaltered thick, scattering specimens. Images of controlled scattering phantoms, blood in collection bags, cerebral organoids and freshly excised whole mouse brains are presented to validate the approach.
TL;DR: A rapidly tunable dual-output all-fiber light source for coherent Raman imaging, based on a dispersively matched mode-locked laser pumping a parametric oscillator, showing a reliable operation despite mechanical shocks and being well suited for operation in a mobile cart is presented.
Abstract: We present a rapidly tunable dual-output all-fiber light source for coherent Raman imaging, based on a dispersively matched mode-locked laser pumping a parametric oscillator. Output pump and Stokes pulses with a maximal power of 170 and 400 mW, respectively, and equal durations of 7 ps could be generated. The tuning mechanism required no mechanical delay line, enabling all-electronic arbitrary wavelength switching across more than 2700 cm−1 in less than 5 ms. The compact setup showed a reliable operation despite mechanical shocks of more than 25 m/s2 and is, thus, well suited for operation in a mobile cart. Imaging mouse and human skin tissue with both the portable light source and a commercial laboratory-bound reference system yielded qualitatively equal results and verified the portable light source being well suited for coherent Raman microscopy.
TL;DR: Application of compressional optical coherence elastography (OCE) for delineation of tumor and peri-tumoral tissue with simultaneous assessment of morphological/molecular subtypes of breast cancer is reported.
Abstract: Application of compressional optical coherence elastography (OCE) for delineation of tumor and peri-tumoral tissue with simultaneous assessment of morphological/molecular subtypes of breast cancer is reported. The approach is based on the ability of OCE to quantitatively visualize stiffness of studied samples and then to perform a kind of OCE-based biopsy by analyzing elastographic B-scans that have sizes ~several millimeters similarly to bioptates used for “gold-standard” histological examinations. The method relies on identification of several main tissue constituents differing in their stiffness in the OCE scans. Initially the specific stiffness ranges for the analyzed tissue components (adipose tissue, fibrous and hyalinized tumor stroma, lymphocytic infiltrate and agglomerates of tumor cells) are determined via comparison of OCE and morphological/molecular data. Then assessment of non-tumor/tumor regions and tumor subtypes is made based on percentage of pixels with different characteristic stiffness (“stiffness spectrum”) in the OCE image, also taking into account spatial localization of different-stiffness regions. Examples of high contrast among benign (or non-invasive) and several subtypes of invasive breast tumors in terms of their stiffness spectra are given.
TL;DR: This work quantifies the mechanical properties of the extra-cellular matrix (ECM) in live zebrafish using Brillouin microscopy and finds the ECM to be ~500 nm thick, and in very good agreement with electron microscopy quantification.
Abstract: In this work, we quantify the mechanical properties of the extra-cellular matrix (ECM) in live zebrafish using Brillouin microscopy. Optimization of the imaging conditions and parameters, combined with careful spectral analysis, allows us to resolve the thin ECM and distinguish its Brillouin frequency shift, a proxy for mechanical properties, from the surrounding tissue. High-resolution mechanical mapping further enables the direct measurement of the thickness of the ECM label-free and in-vivo. We find the ECM to be ~500 nm thick, and in very good agreement with electron microscopy quantification. Our results open the door for future studies that aim to investigate the role of ECM mechanics for zebrafish morphogenesis and axis elongation.
TL;DR: A convolutional neural network is developed, MEDnet-V2, to distinguish NPA from signal reduction artifacts in 6×6 mm2 OCTA and achieves strong specificity and sensitivity for NPA detection across a wide range of DR severity and scan quality.
Abstract: The capillary nonperfusion area (NPA) is a key quantifiable biomarker in the evaluation of diabetic retinopathy (DR) using optical coherence tomography angiography (OCTA). However, signal reduction artifacts caused by vitreous floaters, pupil vignetting, or defocus present significant obstacles to accurate quantification. We have developed a convolutional neural network, MEDnet-V2, to distinguish NPA from signal reduction artifacts in 6×6 mm2 OCTA. The network achieves strong specificity and sensitivity for NPA detection across a wide range of DR severity and scan quality.
TL;DR: This manuscript describes the design and development of a non-contact, handheld probe optimized for OCT angiography that features a novel diverging light on the scanner optical design that provides improved optical performance over traditional OCT scanner designs.
Abstract: OCT angiography is a functional extension of OCT that allows for non-invasive imaging of retinal microvasculature. However, most current OCT angiography systems are tabletop systems that are typically used for imaging compliant, seated subjects. These systems cannot be readily applied for imaging important patient populations such as bedridden patients, patients undergoing surgery in the operating room, young children in the clinic, and infants in the intensive care nursery. In this manuscript, we describe the design and development of a non-contact, handheld probe optimized for OCT angiography that features a novel diverging light on the scanner optical design that provides improved optical performance over traditional OCT scanner designs. Unlike most handheld OCT probes, which are designed to be held by the side of the case or by a handle, the new probe was optimized for ergonomics of supine imaging where imagers prefer to hold the probe by the lens tube. The probe's design also includes an adjustable brace that gives the operator a point of contact closer to the center of mass of the probe, reducing the moment of inertia around the operator's fingers, facilitating stabilization, and reducing operator fatigue. The probe supports high-speed imaging using a 200 kHz swept source OCT engine, has a motorized stage that provides + 10 to -10 D refractive error correction and weighs 700g. We present initial handheld OCT angiography images from healthy adult volunteers, young children during exams under anesthesia, and non-sedated infants in the intensive care nursery. To the best of our knowledge, this represents the first reported use of handheld OCT angiography in non-sedated infants, and the first handheld OCT angiography images which show the clear delineation of key features of the retinal capillary complex including the foveal avascular zone, peripapillary vasculature, the superficial vascular complex, and the deep vascular complex.
TL;DR: The optimal design scheme with the optimally determined mIDT scheme is analyzed, enabling hardware-limited 4Hz acquisition rates enabling 3D refractive index distribution recovery on live Caenorhabditis elegans worms and embryos as well as epithelial buccal cells.
Abstract: Intensity diffraction tomography (IDT) provides quantitative, volumetric refractive index reconstructions of unlabeled biological samples from intensity-only measurements. IDT is scanless and easily implemented in standard optical microscopes using an LED array but suffers from large data requirements and slow acquisition speeds. Here, we develop multiplexed IDT (mIDT), a coded illumination framework providing high volume-rate IDT for evaluating dynamic biological samples. mIDT combines illuminations from an LED grid using physical model-based design choices to improve acquisition rates and reduce dataset size with minimal loss to resolution and reconstruction quality. We analyze the optimal design scheme with our mIDT framework in simulation using the reconstruction error compared to conventional IDT and theoretical acquisition speed. With the optimally determined mIDT scheme, we achieve hardware-limited 4Hz acquisition rates enabling 3D refractive index distribution recovery on live Caenorhabditis elegans worms and embryos as well as epithelial buccal cells. Our mIDT architecture provides a 60 × speed improvement over conventional IDT and is robust across different illumination hardware designs, making it an easily adoptable imaging tool for volumetrically quantifying biological samples in their natural state.
TL;DR: It is demonstrated that live cells are ~80 times less susceptible to the 660 nm incident light compared to 532 nm light, which overall allows Brillouin imaging of up to more than 30 times higher SNR and enables BrillouIn imaging of live biological samples with improved accuracy, higher speed and/or larger fields of views with denser sampling.
Abstract: In Brillouin microscopy, absorption-induced photodamage of incident light is the primary limitation on signal-to-noise ratio in many practical scenarios. Here we show that 660 nm may represent an optimal wavelength for Brillouin microscopy as it offers minimal absorption-mediated photodamage at high Brillouin scattering efficiency and the possibility to use a pure and narrow laser line from solid-state lasing medium. We demonstrate that live cells are ~80 times less susceptible to the 660 nm incident light compared to 532 nm light, which overall allows Brillouin imaging of up to more than 30 times higher SNR. We show that this improvement enables Brillouin imaging of live biological samples with improved accuracy, higher speed and/or larger fields of views with denser sampling.
TL;DR: The immunohistochemical and confocal data clearly demonstrate the significant reduction of deposition of Aβ plaques in mice after tPBM vs. untreated animals, and open breakthrough strategies for a non-pharmacological therapy of Alzheimer's disease.
Abstract: In this pilot study, we analyzed effects of transcranial photobiomodulation (tPBM, 1267 nm, 32 J/cm2) on clearance of beta-amyloid (Aβ) from the mouse brain. The immunohistochemical and confocal data clearly demonstrate the significant reduction of deposition of Aβ plaques in mice after tPBM vs. untreated animals. The behavior tests showed that tPBM improved the cognitive, memory and neurological status of mice with Alzheimer’s disease (AD). Using of our original method based on optical coherence tomography (OCT) analysis of clearance of gold nanorods (GNRs) from the brain, we proposed possible mechanism underlying tPBM-stimulating effects on clearance of Aβ via the lymphatic system of the brain and the neck. These results open breakthrough strategies for a non-pharmacological therapy of Alzheimer’s disease and clearly demonstrate that tPBM might be a promising therapeutic target for preventing or delaying Alzheimer’s disease.
TL;DR: This review summarizes recent data and topical problems in nanomedicine that are related to the use of variously sized, shaped, and structured GNPs.
Abstract: Functionalized gold nanoparticles (GNPs) with controlled geometrical and optical properties have been the subject of intense research and biomedical applications. This review summarizes recent data and topical problems in nanomedicine that are related to the use of variously sized, shaped, and structured GNPs. We focus on three topical fields in current nanomedicine: (1) use of GNP-based nanoplatforms for the targeted delivery of anticancer and antimicrobial drugs and of genes; (2) GNP-based cancer immunotherapy; and (3) combined chemo-, immuno-, and phototherapy. We present a summary of the available literature data and a short discussion of future work.
TL;DR: A setup for multiplexed distributed optical fiber sensors capable of resolving temperature distribution in thermo-therapies, with a spatial resolution of 2.5 mm over multiple fibers interrogated simultaneously, validated for the planar measurement of temperature profiles in ex vivo radiofrequency ablation.
Abstract: We propose a setup for multiplexed distributed optical fiber sensors capable of resolving temperature distribution in thermo-therapies, with a spatial resolution of 2.5 mm over multiple fibers interrogated simultaneously. The setup is based on optical backscatter reflectometry (OBR) applied to optical fibers having backscattered power significantly larger than standard fibers (36.5 dB), obtained through MgO doping. The setup is based on a scattering-level multiplexing, which allows interrogating all the sensing fibers simultaneously, thanks to the fact that the backscattered power can be unambiguously associated to each fiber. The setup has been validated for the planar measurement of temperature profiles in ex vivo radiofrequency ablation, obtaining the measurement of temperature over a surface of 96 total points (4 fibers, 8 sensing points per cm2). The spatial resolution obtained for the planar measurement allows extending distributed sensing to surface, or even three-dimensional, geometries performing temperature sensing in the tissue with millimeter resolution in multiple dimensions.
TL;DR: This work reports on a system that can acquire essentially crosstalk-free volumes of the retina by using a fast deformable membrane, which enables the visualization of choroids and a clear delineation of the retinal layers that is not possible with conventional FD-FF-OCT.
Abstract: Fourier-domain full-field optical coherence tomography (FD-FF-OCT) is currently the fastest volumetric imaging technique that is able to generate a single 3-D volume of retina in less than 9 ms, corresponding to a voxel rate of 7.8 GHz. FD-FF-OCT is based on a fast camera, a rapidly tunable laser source, and Fourier-domain signal detection. However, crosstalk appearing due to multiply scattered light corrupts images with the speckle pattern, and therefore, lowers image quality. Here, for the first time, we report on a system that can acquire essentially crosstalk-free volumes of the retina by using a fast deformable membrane. It enables the visualization of choroids and a clear delineation of the retinal layers that is not possible with conventional FD-FF-OCT.
TL;DR: A deep network is presented that extracts continuous, smooth, and topology-guaranteed surfaces and MMEs with better accuracy than the state-of-the-art method and learns shape priors automatically during training rather than being hard-coded as in graph methods.
Abstract: Optical coherence tomography (OCT) is a noninvasive imaging modality that can be used to obtain depth images of the retina. Patients with multiple sclerosis (MS) have thinning retinal nerve fiber and ganglion cell layers, and approximately 5% of MS patients will develop microcystic macular edema (MME) within the retina. Segmentation of both the retinal layers and MME can provide important information to help monitor MS progression. Graph-based segmentation with machine learning preprocessing is the leading method for retinal layer segmentation, providing accurate surface delineations with the correct topological ordering. However, graph methods are time-consuming and they do not optimally incorporate joint MME segmentation. This paper presents a deep network that extracts continuous, smooth, and topology-guaranteed surfaces and MMEs. The network learns shape priors automatically during training rather than being hard-coded as in graph methods. In this new approach, retinal surfaces and MMEs are segmented together with two cascaded deep networks in a single feed forward propagation. The proposed framework obtains retinal surfaces (separating the layers) with sub-pixel surface accuracy comparable to the best existing graph methods and MMEs with better accuracy than the state-of-the-art method. The full segmentation operation takes only ten seconds for a 3D volume.
TL;DR: This work presents a method to co-optimize how a sample is illuminated in a microscope, along with a pipeline to automatically classify the resulting image, using a deep neural network, to increase the speed and accuracy of automated image classification.
Abstract: Since its invention, the microscope has been optimized for interpretation by a human observer. With the recent development of deep learning algorithms for automated image analysis, there is now a clear need to re-design the microscope's hardware for specific interpretation tasks. To increase the speed and accuracy of automated image classification, this work presents a method to co-optimize how a sample is illuminated in a microscope, along with a pipeline to automatically classify the resulting image, using a deep neural network. By adding a "physical layer" to a deep classification network, we are able to jointly optimize for specific illumination patterns that highlight the most important sample features for the particular learning task at hand, which may not be obvious under standard illumination. We demonstrate how our learned sensing approach for illumination design can automatically identify malaria-infected cells with up to 5-10% greater accuracy than standard and alternative microscope lighting designs. We show that this joint hardware-software design procedure generalizes to offer accurate diagnoses for two different blood smear types, and experimentally show how our new procedure can translate across different experimental setups while maintaining high accuracy.
TL;DR: By combining static, dynamic and fluorescence contrasts, this work shows that by combining label-free high-resolution imaging of the retina and anterior eye with temporal resolution from milliseconds to several hours, allowing us to probe biological activity at subcellular scales inside 3D bulk tissue.
Abstract: We describe recent technological progress in multimodal en face full-field optical coherence tomography that has allowed detection of slow and fast dynamic processes in the eye. We show that by combining static, dynamic and fluorescence contrasts we can achieve label-free high-resolution imaging of the retina and anterior eye with temporal resolution from milliseconds to several hours, allowing us to probe biological activity at subcellular scales inside 3D bulk tissue. Our setups combine high lateral resolution over a large field of view with acquisition at several hundreds of frames per second which make it a promising tool for clinical applications and biomedical studies. Its contactless and non-destructive nature is shown to be effective for both following in vitro sample evolution over long periods of time and for imaging of the human eye in vivo.
TL;DR: Three advances to further the performance of visible light OCT in the human retina are introduced, including a grating light valve spatial light modulator (GLV-SLM) spectral shaping stage to modify the source spectrum and a novel, Fourier transform-free, software axial motion tracking algorithm with fast, magnetically actuated stage.
Abstract: Visible light optical coherence tomography (OCT) theoretically provides finer axial resolution than near-infrared OCT for a given wavelength bandwidth. To realize this potential in the human retina in vivo, the unique technical challenges of visible light OCT must be addressed. We introduce three advances to further the performance of visible light OCT in the human retina. First, we incorporate a grating light valve spatial light modulator (GLV-SLM) spectral shaping stage to modify the source spectrum. This enables comfortable subject alignment with a red light spectrum, and image acquisition with a broad “white light” spectrum, shaped to minimize sidelobes. Second, we develop a novel, Fourier transform-free, software axial motion tracking algorithm with fast, magnetically actuated stage to maintain near-optimal axial resolution and sensitivity in the presence of eye motion. Third, we implement spatially dependent numerical dispersion compensation for the first time in the human eye in vivo. In vivo human retinal OCT images clearly show that the inner plexiform layer consists of 3 hyper-reflective bands and 2 hypo-reflective bands, corresponding with the standard anatomical division of the IPL. Wavelength-dependent images of the outer retina suggest that, beyond merely improving the axial resolution, shorter wavelength visible light may also provide unique advantages for visualizing Bruch’s membrane.
TL;DR: Characterized by THz time domain spectroscopic absorption quantification measurements with different concentrations of bovine serum albumin (BSA), the proposed sensor exhibits promising applications in microfluidic biosensing.
Abstract: We theoretically and experimentally demonstrate a label-free terahertz biosensor with ultrahigh sensitivity and distinctive discretion. By constructing a metal-air-metal (MAM) metamaterial perfect absorber (MPA) with a metallic paired-ring resonator array, a hollow microfluidic channel, and a backed reflector, a novel dual-band absorptive sensing platform is proposed in the THz range. The near field coupling by dipole-induced trapped modes and the magnetic momentum caused a vertical to transverse power flux that dramatically enhanced the electromagnetic field on top of the metasurface and in the microfluidic channel, respectively. Both the resonant modes exhibit perfect absorption and produce ultrahigh normalized sensitivities of 0.47/RIU (refractive index unit, RIU) and 0.51/RIU at 0.76 THz and 1.28 THz, respectively. Compared with conventional microfluidic sensors, the salient advantages of our design are the perfect spatial overlap for light-matter interaction and polarization insensitivity. Characterized by THz time domain spectroscopic absorption quantification measurements with different concentrations of bovine serum albumin (BSA), the proposed sensor exhibits promising applications in microfluidic biosensing.
TL;DR: This work applies a recently developed quantitative method, called ultraviolet hyperspectral interferometry (UHI), to characterize the dispersion and absorbing properties of various important biomolecules, including hemoglobin, beta nicotinamide adenine dinucleotide, elastin, collagen, cytochrome c, tryptophan and DNA.
Abstract: Owing to the high precision and sensitivity of optical systems, there is an increasing demand for optical methods that quantitatively characterize the physical and chemical properties of biological samples. Information extracted from such quantitative methods, through phase and/or amplitude variations of light, can be crucial in the diagnosis, treatment and study of disease. In this work we apply a recently developed quantitative method, called ultraviolet hyperspectral interferometry (UHI), to characterize the dispersion and absorbing properties of various important biomolecules. Our system consists of (1) a broadband light source that spans from the deep-UV to the visible region of the spectrum, and (2) a Mach–Zehnder interferometer to gain access to complex optical properties. We apply this method to characterize (and tabulate) the dispersive and absorptive properties of hemoglobin, beta nicotinamide adenine dinucleotide (NAD), flavin adenine dinucleotide (FAD), elastin, collagen, cytochrome c, tryptophan and DNA. Our results shed new light on the complex properties of important biomolecules.
TL;DR: High sensitivity and specificity were accomplished, which offers potential for hyperspectral imaging as a margin assessment technique to improve surgical outcome.
Abstract: Complete tumor removal during breast-conserving surgery remains challenging due to the lack of optimal intraoperative margin assessment techniques. Here, we use hyperspectral imaging for tumor detection in fresh breast tissue. We evaluated different wavelength ranges and two classification algorithms; a pixel-wise classification algorithm and a convolutional neural network that combines spectral and spatial information. The highest classification performance was obtained using the full wavelength range (450-1650 nm). Adding spatial information mainly improved the differentiation of tissue classes within the malignant and healthy classes. High sensitivity and specificity were accomplished, which offers potential for hyperspectral imaging as a margin assessment technique to improve surgical outcome.