Optical Coherence Tomography Image Analysis Research Articles

Deep attention for enhanced OCT image analysis in clinical retinal diagnosis

AbstractRetinal illnesses such as age-related macular degeneration (AMD) and diabetic maculopathy pose serious risks to vision in the developed world. The diagnosis and assessment of these disorders have undergone revolutionary change with the development of optical coherence tomography (OCT). This study proposes a novel method for improving clinical precision in retinal disease diagnosis by utilizing the strength of Attention-Based DenseNet, a deep learning architecture with attention processes. For model building and evaluation, a dataset of 84495 high-resolution OCT images divided into NORMAL, CNV, DME, and DRUSEN classes was used. Data augmentation techniques were employed to enhance the model's robustness. The Attention-Based DenseNet model achieved a validation accuracy of 0.9167 with a batch size of 32 and 50 training epochs. This discovery presents a promising route for more precise and speedy identification of retinal illnesses, ultimately enhancing patient care and outcomes in clinical settings by integrating cutting-edge technology with powerful neural network architectures.

Optical Coherence Tomography Image Analysis for Detection of Alzheimer’s Disease: A Comprehensive Structured Review

Optical Coherence Tomography (OCT) has emerged as a promising non-invasive imaging modality for the early detection of Alzheimer’s Disease (AD). This systematic literature review aims to consolidate current research on OCT image analysis for AD detection, addressing the growing need for early and accurate diagnostic tools. Despite the advances in neuroimaging, early diagnosis of AD remains challenging due to its asymptomatic nature in initial stages and the invasiveness of traditional methods. To achieve this, we conducted an extensive search of related articles from reputable databases (Scopus and Web of Science), focusing on studies published between 2022-2024. The flow of study was based on PRISMA framework. The database found (n = 29) final primary data. This review was divided into three themes, (1) retinal and ocular biomarkers for AD, (2) optical coherence tomography angiography (OCTA) and imaging techniques, and (3) machine learning and computational approaches for Alzheimer’s disease diagnosis. Key findings include the enlargement of the periarteriole capillary-free zone and changes in retinal nerve fibre layer thickness as potential AD biomarkers. Based on the review, the implementation of image analysis for OCT images have shown substantial potential for AD detection. By evaluating the past studies, gaps for the current research were discovered including the need for larger, more diverse cohorts and longitudinal studies to validate these biomarkers. In summary, detection of AD is possible through thorough OCT image analysis, but further research could be suggested to enhance its clinical applicability and reliability.

Fractal geometry of culprit coronary plaque images within optical coherence tomography in patients with acute coronary syndrome vs stable angina pectoris.

The main cause of acute coronary syndrome (ACS) is plaque rupture and thrombus formation. However, it has not been fairly successful to identify vulnerable plaque to rupture using conventional parameters of intravascular imaging modalities. Fractal analysis is one of the mathematical models to examine geometrical features of picture image using a specific parameter called as fractal dimension (FD) which suggests geometric complexity of the image. This study examined FD of the optical coherence tomography (OCT)-derived images of the culprit plaque in patients with ACS vs stable angina pectoris (SAP) to evaluate the feasibility of FD for identifying vulnerable coronary plaques prone to provoke ACS distinguished from stable plaques only provoking SAP. We examined 65 cases (34 ACS patients, 31 SAP patients) in which the culprit lesion was imaged by OCT before percutaneous coronary intervention in patients with ACS and SAP. The culprit plaque lesion in the ACS group had a significantly larger mean lipid arc (203.8 ± 39.4° vs 152.3 ± 34.5°, p < 0.001) and a larger lipid plaque length (12.6 ± 5.1mm vs 7.7 ± 2.7mm, p < 0.001) and a thinner fibrous cap thickness (75.3 ± 22.3μm vs 134.8 ± 53.2μm, p < 0.001) than those in the SAP group. The prevalence of OCT-derived macrophage infiltration (Mph) in the entire culprit coronary vessel as well as that of the OCT-derived thin-cap fibroatheroma (TCFA) at the culprit lesion were significantly greater in the ACS group than those in the SAP group, respectively (Mph: 61.8% vs 35.5%, p = 0.048; TCFA: 44.1% vs 6.4%, p < 0.001). The FD of culprit plaque in the ACS group was significantly greater than in the SAP group (2.401 ± 0.073 vs 2.341 ± 0.051, p < 0.001). In multivariate regression analysis, the presence of Mph was a significant determinant of FD (regression coefficient estimate 0.049, CI 0.018-0.079, p = 0.002). The FD of OCT-derived image of culprit coronary plaque in the ACS group was significantly greater than that in the SAP group, indicating that the culprit plaque in ACS were structurally more complex. Therefore, fractal analysis of coronary OCT images might be clinically useful for identifying coronary plaques prone to provoke ACS.

The effect of platelet-rich plasma and sodium alginate hydrogel on corneal wound healing after corneal alkali burns in rats with computer-assisted anterior segment optical coherence tomography image analysis

Our objective was to determine the effect of a semi-synthetic sodium alginate hydrogel and its combination with platelet-rich plasma (PRP) on histopathological, biochemical, clinical, and anterior segment optical coherence tomography (AS-OCT) data. Alkali chemical burn of the cornea was induced. Injured rats were randomly divided into five equal groups and topically treated with phosphate-buffered saline (sham), platelet-rich plasma (PRP), 0.5% sodium citrate, a semi-synthetic sodium alginate hydrogel, or a combination of PRP and hydrogel (combined group) three times daily. The degree of corneal opacity (CO), corneal epithelial staining (CES), percentage of corneal epithelial defects (CEDP), degree of ciliary hyperemia (CH), neovascularization size (NVS), and extent of neovascularization (NVE) were evaluated. AS-OCT was performed at nine days, and then rats were sacrificed. Histological examination and enzyme-linked immunosorbent assays were performed to detect the concentrations of IL-1β and MMP-9 in the cornea. There were no significant differences between the groups regarding CEDP, CO, CES, CH, NVS, or NVE on the first day after corneal alkali burn injury (p > 0,05). At the last examination, CO was significantly lower in the PRP group than in the sham group (p = 0,044), while the CO concentrations were similar in terms of NVS (p > 0,05). Similarly, in terms of tissue MMP-9 levels, there were no significant differences between groups (p > 0,05). However, there was a significant difference in tissue IL-1β levels between the groups (p < 0,001). In the PRP and combined groups, the level of IL-1β was significantly lower than that in the sham group (p = 0,043 and p = 0,036, respectively). There was a significant difference in epithelial necrosis between the PRP, and it was the lowest in the combined group (p = 0,003). Epithelial thickness was highest in the combined group (p = 0,002). CEDP was significantly different at the last visit between the groups (p = 0.042). The fastest epithelial closing rate was observed for the combined group (p = 0,026). There was a significant negative correlation between tissue MMP-9 levels and corneal solidity and between tissue MMP-9 levels and the corneal area according to the AS-OCT measurements (p = 0,012 and p = 0,027, respectively). When used alone, topical hydrogel application did not significantly enhance the healing of corneal wounds. However, when combined with PRP, it leads to an increased rate of epithelial closure and neovascularization. This combination did not exacerbate inflammation or corneal opacity compared to PRP alone. The anticoagulant citrate solution in the PRP tube did not prove effective. The synergistic use of PRP and hydrogel could enhance epithelial thickness and reduce epithelial necrosis. The use of new parameters for corneal wound healing assessment was facilitated through AS-OCT image processing.

Polarization-resolved analysis of outer retinal bands: investigating ballistic and multiply scattered photons using full-field swept-source optical coherence tomography.

Precise interpretation of the anatomical origins of outer retinal optical coherence tomography (OCT) presents technical challenges owing to the delicate nature of the retina. To address this challenge, our study introduces a novel polarization-sensitive full-field swept-source OCT (FF-SS-OCT) that provides parallel-polarization and cross-polarization OCT measurements, predominantly capturing ballistically reflected photons and multiply scattered photons, respectively. Notably, parallel-polarization OCT unveils layer-like structures more effectively, including the inner plexiform layer (IPL) sub-layers, outer plexiform layer (OPL) sub-layers (in rod-dominant regions), and rod/cone outer segment (OS) tips, compared to cross-polarization OCT, where such sub-layers are not visible. Through a comparative analysis of parallel-polarization and cross-polarization OCT images of the outer retina, we discovered that the 2nd outer retinal OCT band results from contributions from both the ellipsoid zone (EZ) and the inner segment/outer segment (IS/OS) junction. Similarly, the 3rd outer retinal OCT band appears to reflect contributions from both the interdigitation zone (IZ) and photoreceptor OS tips. This polarization-sensitive approach advances our understanding of the origins of outer retinal OCT signals and proposes potential new biomarkers for assessing retinal health and diseases.

Quantifying Changes on OCT in Eyes Receiving Treatment for Neovascular Age-Related Macular Degeneration

PurposeApplication of artificial intelligence (AI) to macular optical coherence tomography (OCT) scans to segment and quantify volumetric change of anatomical and pathological features during intravitreal treatment for neovascular age-related macular degeneration (AMD). DesignRetrospective analysis of OCT images from the Moorfields Eye Hospital AMD Database. Participants2115 eyes from 1883 patients that started anti-vascular endothelial growth factor (anti-VEGF) treatment between 1st June 2012 and 30th June 2017. MethodsThe Moorfields Eye Hospital neovascular AMD database was queried for first and second eyes that received anti-VEGF treatment and had an OCT scan at baseline and at 12 months. Follow-up scans were input into the AI system to derive volumetric outputs for the following variables: intraretinal fluid (IRF), subretinal fluid (SRF), pigment epithelial detachment (PED), subretinal hyperreflective material (SHRM), hyperreflective foci (HRF), neurosensory retina (NSR) and retinal pigment epithelium (RPE). Volumes were studied at different time points and comparisons made with respect to baseline volume groups. Cross-sectional comparisons between time points were conducted using Mann-Whitney U test. ResultsMean volumes of analysed features decreased significantly from baseline to both four and 12 months, in both first and second-treated eyes. Pathological features that reflect exudation, including pure fluid components (IRF and SRF) and those with a mixture of fluid and fibrovascular tissue (PED and SHRM), displayed similar response to treatment over 12 months. Mean PED and SHRM volumes showed less pronounced but also substantial decreases over the first two months, reaching a plateau post loading phase, and minimal change to 12 months. Both NSR and RPE volumes showed gradual reductions over time, and not as substantial as exudative features. Conclusion:We report the results of a quantitative analysis of change in retinal segmented features over time, enabled by an AI segmentation system. Cross-sectional analysis at multiple time points demonstrated significant associations between baseline OCT-derived segmented features and the volume of biomarkers at follow-up. Demonstrating how certain OCT biomarkers progress with treatment and the impact of pre-treatment retinal morphology on different structural volumes may provide novel insights into disease mechanisms and aid personalization of care. Data will be made public for future studies.

Assessing the efficacy of 2D and 3D CNN algorithms in OCT-based glaucoma detection

Glaucoma is a progressive neurodegenerative disease characterized by the gradual degeneration of retinal ganglion cells, leading to irreversible blindness worldwide. Therefore, timely and accurate diagnosis of glaucoma is crucial, enabling early intervention and facilitating effective disease management to mitigate further vision deterioration. The advent of optical coherence tomography (OCT) has marked a transformative era in ophthalmology, offering detailed visualization of the macula and optic nerve head (ONH) regions. In recent years, both 2D and 3D convolutional neural network (CNN) algorithms have been applied to OCT image analysis. While 2D CNNs rely on post-prediction aggregation of all B-scans within OCT volumes, 3D CNNs allow for direct glaucoma prediction from the OCT data. However, in the absence of extensively pre-trained 3D models, the comparative efficacy of 2D and 3D-CNN algorithms in detecting glaucoma from volumetric OCT images remains unclear. Therefore, this study explores the efficacy of glaucoma detection through volumetric OCT images using select state-of-the-art (SOTA) 2D-CNN models, 3D adaptations of these 2D-CNN models with specific weight transfer techniques, and a custom 5-layer 3D-CNN-Encoder algorithm. The performance across two distinct datasets is evaluated, each focusing on the macula and the ONH, to provide a comprehensive understanding of the models’ capabilities in identifying glaucoma. Our findings demonstrate that the 2D-CNN algorithm consistently provided robust results compared to their 3D counterparts tested in this study for glaucoma detection, achieving AUC values of 0.960 and 0.943 for the macular and ONH OCT test images, respectively. Given the scarcity of pre-trained 3D models trained on extensive datasets, this comparative analysis underscores the overall utility of 2D and 3D-CNN algorithms in advancing glaucoma diagnostic systems in ophthalmology and highlights the potential of 2D algorithms for volumetric OCT image-based glaucoma detection.

Discriminating Glioblastoma from Normal Brain Tissue In Vivo Using Optical Coherence Tomography and Angiography: A Texture and Microvascular Analysis Approach

Brain tumors arise from abnormal cell growth in the brain. Glioblastoma, the most common and aggressive type, poses significant challenges for identification during surgery. The primary goal of this study is to identify and differentiate normal brain tissue from glioblastoma tissue using optical coherence tomography (OCT) and OCT angiography (OCTA). These techniques offer a non-invasive way to analyze the morphological and microvascular alternations associated with glioblastoma in an animal model. To monitor the changes in morphology and vascular distribution of brain tissue as glioblastoma tumors grow, time-series OCT and OCTA results were collected for comparison. Texture analysis of OCT images was proposed using the gray-level co-occurrence matrix (GLCM), from which homogeneity and variance were calculated as discriminative parameters. Additionally, OCTA was used to assess microvascular characteristics, including vessel diameter, density, and fractal dimension. The findings demonstrate that the proposed methods can effectively distinguish between normal and cancerous brain tissue in vivo.

Unveiling Key Pathological Indicators for Disease Progression in Vogt Koyanagi Harada Disease and Sympathetic Ophthalmia Through Advanced Choroidal Volume Analysis

ABSTRACT Purpose To evaluate the association between quantitative parameters derived from volume analysis of optical coherence tomography (OCT) data and disease worsening in Vogt-Koyanagi-Harada disease (VKHD) and sympathetic ophthalmia (SO). Methods This retrospective study, conducted at Osaka University Hospital, employed swept-source OCT scans from patients diagnosed with VKHD or SO between October 2012 and January 2021. The choroidal vessel structure was segmented and visualized in three dimensions, generating quantitative vessel volume maps. Region-specific choroidal vessel volume (CVV), choroidal volume (CV), and vessel index (VI) were scrutinized for their potential correlation with disease severity. Results Thirty-five eyes of 18 VKHD and 2 SO patient (8 females, 10 males) were evaluated. OCT-derived CVV maps revealed regional CV alterations in VKHD and SO patients. Two parameters, i.e. CV at 3- and 6-month follow-ups (p = 0.044, p = 0.040, respectively, with area under the ROC curve of 0.70) and CVV at 6 months (p = 0.046, area under the ROC curve of 0.71), were significantly higher in recurrent VKHD and SO compared to effectively treated cases. Conclusions The volume analysis of OCT images facilitates a three-dimensional visualization of choroidal alterations, which may serve as a reflection of disease severity in VKHD and SO patients. Furthermore, noninvasive initial CVV or CV measurements may serve as potential biomarkers for predicting disease recurrence in VKHD and SO.

Automated Identification of Retinal Disorders by Deep Learning with Optical Coherence Tomography Pictures

A crucial problem in computer-aided diagnosis (CAD) for medical applications is the classification of retinal disorders. This paper addresses the 4-class classification problem in optical coherence tomography image analysis (OCT) to automatically identify choroidal neovascularization (CNV), diabetic macular edema (DME), DRUSEN, and NORMAL. Retinal OCT pictures were identified by the proposed classification approach using a set of four classification model instances, each based on an upgraded residual neural network (ResNet50).At the patient level, the experiment employs a 10-fold cross-validation process that is founded on the growth of the OCT retinal picture collection. According to the study's findings, multiple ResNet50 concatenation is a helpful method when the availability of pictures used in medicine is scarce. In order to better understand the decision-making process, we also ran a bottleneck test and a qualitative assessment of the model predictions.

Deep-learning based analysis of intracoronary optical coherence tomography images: detection and characterization of lipid plaques

Abstract Background Intracoronary optical coherence tomography (OCT) enables in vivo detection and characterization of atherosclerotic disease. However, manual assessment of OCT images is time-consuming and subject to intra- and interobserver variability. Therefore, automated and trustworthy methods for plaque assessment are warranted. Purpose To develop and validate an artificial intelligence based algorithm for detection and characterization of lipid plaques on OCT images. Methods From the prospective, observational PECTUS-obs study, we studied a representative subset of OCT pullbacks performed in fractional flow reserve-negative non-culprit lesions in patients presenting with myocardial infarction. Manual segmentation of cross-sectional frames (e.g. lumen, tunica intima including the fibrous cap, tunica media and lipid pools) was performed by trained experts. Pullbacks were randomly divided into a training and test set. For automated segmentation and analysis, a two-dimensional no-new U-shaped neural network (nnUNet) was constructed with 5-fold cross validation. To test the diagnostic performance of the nnUnet for detection of lipid plaques, sensitivity, specificity and Cohen’s kappa were calculated. As for plaque characterization, the lipid arc and minimal fibrous cap thickness were measured manually in the test set on each frame containing a lipid plaque. Values of lipid arc and minimal fibrous cap thickness obtained following automated assessment were compared to manual analysis using the intraclass correlation coefficient (ICC) for absolute agreement. Results In total, 1215 frames were used for training and 162 frames for testing of the nnUNet. Lipid plaques were present in 77 frames (47.5%) in the test set. Substantial agreement (κ=0.79) for detection of a lipid plaque was achieved with a sensitivity and specificity of 96.1% and 83.5%, respectively. As for plaque characterization, we found good reliability for lipid arc assessment (ICC 0.88, mean difference 6±32º) and moderate reliability for assessment of the minimal fibrous cap thickness (ICC 0.54, mean difference 23±140µm). Conclusion We developed a deep learning algorithm for automated analysis of intracoronary OCT images, that was able to accurately detect lipid plaques. Moreover, the algorithm showed promising results in terms of plaque characterization. With further refinements, the developed algorithm has the potential to reduce the time required for OCT interpretation, ultimately enabling more efficient, real-time use of this imaging modality in clinical practice.

Fractional Flow Reserve–Negative High-Risk Plaques and Clinical Outcomes After Myocardial Infarction

Even after fractional flow reserve (FFR)-guided complete revascularization, patients with myocardial infarction (MI) have high rates of recurrent major adverse cardiovascular events (MACE). These recurrences may be caused by FFR-negative high-risk nonculprit lesions. To assess the association between optical coherence tomography (OCT)-identified high-risk plaques of FFR-negative nonculprit lesions and occurrence of MACE in patients with MI. PECTUS-obs (Identification of Risk Factors for Acute Coronary Events by OCT After STEMI [ST-segment elevation MI] and NSTEMI [non-STEMI] in Patients With Residual Non-flow Limiting Lesions) is an international, multicenter, prospective, observational cohort study. In patients presenting with MI, OCT was performed on all FFR-negative (FFR > 0.80) nonculprit lesions. A high-risk plaque was defined containing at least 2 of the following prespecified criteria: (1) a lipid arc at least 90°, (2) a fibrous cap thickness less than 65 μm, and (3) either plaque rupture or thrombus presence. Patients were enrolled from December 14, 2018, to September 15, 2020. Data were analyzed from December 2, 2022, to June 28, 2023. The primary end point of MACE, a composite of all-cause mortality, nonfatal MI, or unplanned revascularization, at 2-year follow-up was compared in patients with and without a high-risk plaque. A total of 438 patients were enrolled, and OCT findings were analyzable in 420. Among included patients, mean (SD) age was 63 (10) years, 340 (81.0) were men, and STEMI and non-STEMI were equally represented (217 [51.7%] and 203 [48.3%]). A mean (SD) of 1.17 (0.42) nonculprit lesions per patient was imaged. Analysis of OCT images revealed at least 1 high-risk plaque in 143 patients (34.0%). The primary end point occurred in 22 patients (15.4%) with a high-risk plaque and 23 of 277 patients (8.3%) without a high-risk plaque (hazard ratio, 1.93 [95% CI, 1.08-3.47]; P = .02), primarily driven by more unplanned revascularizations in patients with a high-risk plaque (14 of 143 [9.8%] vs 12 of 277 [4.3%]; P = .02). Among patients with MI and FFR-negative nonculprit lesions, the presence of a high-risk plaque is associated with a worse clinical outcome, which is mainly driven by a higher number of unplanned revascularizations. In a population with a high recurrent event rate despite physiology-guided complete revascularization, these results call for research on additional pharmacological or focal treatment strategies in patients harboring high-risk plaques.

Deep Learning Applications to Classification and Detection of Age-Related Macular Degeneration on Optical Coherence Tomography Imaging: A Review.

Age-related macular degeneration (AMD) is one of the leading causes of blindness in the elderly, more commonly in developed countries. Optical coherence tomography (OCT) is a non-invasive imaging device widely used for the diagnosis and management of AMD. Deep learning (DL) uses multilayered artificial neural networks (NN) for feature extraction, and is the cutting-edge technique for medical image analysis for diagnostic and prognostication purposes. Application of DL models to OCT image analysis has garnered significant interest in recent years. In this review, we aimed to summarize studies focusing on DL models used in classification and detection of AMD. Additionally, we provide a brief introduction to other DL applications in AMD, such as segmentation, prediction/prognostication, and models trained on multimodal imaging.

Deep learning in optical coherence tomography: Where are the gaps?

Optical coherence tomography (OCT) is a non-invasive optical imaging modality, which provides rapid, high-resolution and cross-sectional morphology of macular area and optic nerve head for diagnosis and managing of different eye diseases. However, interpreting OCT images requires experts in both OCT images and eye diseases since many factors such as artefacts and concomitant diseases can affect the accuracy of quantitative measurements made by post-processing algorithms. Currently, there is a growing interest in applying deep learning (DL) methods to analyse OCT images automatically. This review summarises the trends in DL-based OCT image analysis in ophthalmology, discusses the current gaps, and provides potential research directions. DL in OCT analysis shows promising performance in several tasks: (1) layers and features segmentation and quantification; (2) disease classification; (3) disease progression and prognosis; and (4) referral triage level prediction. Different studies and trends in the development of DL-based OCT image analysis are described and the following challenges are identified and described: (1) public OCT data are scarce and scattered; (2) models show performance discrepancies in real-world settings; (3) models lack of transparency; (4) there is a lack of societal acceptance and regulatory standards; and (5) OCT is still not widely available in underprivileged areas. More work is needed to tackle the challenges and gaps, before DL is further applied in OCT image analysis for clinical use.

Boundary-enhanced semi-supervised retinal layer segmentation in optical coherence tomography images using fewer labels.

Automatic segmentation of multiple layers in retinal optical coherence tomography (OCT) images is crucial for eye disease diagnosis and treatment. Despite the success of deep learning algorithms, it still remains a challenge due to the blurry layer boundaries and lack of adequate pixel-wise annotations. To tackle these issues, we propose a Boundary-Enhanced Semi-supervised Network (BE-SemiNet) that exploits an auxiliary distance map regression task to improve retinal layer segmentation with scarce labeled data and abundant unlabeled data. Specifically, a novel Unilaterally Truncated Distance Map (UTDM) is firstly introduced to alleviate the class imbalance problem and enhance the layer boundary learning in the regression task. Then for the pixel-wise segmentation and UTDM regression branches, we impose task-level and data-level consistency regularization on unlabeled data to enrich the diversity of unsupervised information and improve the regularization effects. Pseudo supervision is incorporated in consistency regularization to bridge the task prediction spaces for consistency and expand training labeled data. Experiments on two public retinal OCT datasets show that our method can greatly improve the supervised baseline performance with only 5 annotations and outperform the state-of-the-art methods. Since it is difficult and labor-expensive to obtain adequate pixel-wise annotations in practice, our method has a promising application future in clinical retinal OCT image analysis.

On Machine Learning in Clinical Interpretation of Retinal Diseases Using OCT Images.

Optical coherence tomography (OCT) is a noninvasive imaging technique that provides high-resolution cross-sectional retina images, enabling ophthalmologists to gather crucial information for diagnosing various retinal diseases. Despite its benefits, manual analysis of OCT images is time-consuming and heavily dependent on the personal experience of the analyst. This paper focuses on using machine learning to analyse OCT images in the clinical interpretation of retinal diseases. The complexity of understanding the biomarkers present in OCT images has been a challenge for many researchers, particularly those from nonclinical disciplines. This paper aims to provide an overview of the current state-of-the-art OCT image processing techniques, including image denoising and layer segmentation. It also highlights the potential of machine learning algorithms to automate the analysis of OCT images, reducing time consumption and improving diagnostic accuracy. Using machine learning in OCT image analysis can mitigate the limitations of manual analysis methods and provide a more reliable and objective approach to diagnosing retinal diseases. This paper will be of interest to ophthalmologists, researchers, and data scientists working in the field of retinal disease diagnosis and machine learning. By presenting the latest advancements in OCT image analysis using machine learning, this paper will contribute to the ongoing efforts to improve the diagnostic accuracy of retinal diseases.

Nondestructive label-free detection of peritumoral white matter damage using cross-polarization optical coherence tomography

To improve the quality of brain tumor resections, it is important to differentiate zones with myelinated fibers destruction from tumor tissue and normal white matter. Optical coherence tomography (OCT) is a promising tool for brain tissue visualization and in the present study, we demonstrate the ability of cross-polarization (CP) OCT to detect damaged white matter and differentiate it from normal and tumor tissues. The study was performed on 215 samples of brain tissue obtained from 57 patients with brain tumors. The analysis of the obtained OCT data included three stages: 1) visual analysis of structural OCT images; 2) quantitative assessment based on attenuation coefficients estimation in co- and cross-polarizations; 3) building of color-coded maps with subsequent visual analysis. The defining characteristics of structural CP OCT images and color-coded maps were determined for each studied tissue type, and then two classification tests were passed by 8 blinded respondents after a training. Visual assessment of structural CP OCT images allows detecting white matter areas with damaged myelinated fibers and differentiate them from normal white matter and tumor tissue. Attenuation coefficients also allow distinguishing all studied brain tissue types, while it was found that damage to myelinated fibers leads to a statistically significant decrease in the values of attenuation coefficients compared to normal white matter. Nevertheless, the use of color-coded optical maps looks more promising as it combines the objectivity of optical coefficient and clarity of the visual assessment, which leads to the increase of the diagnostic accuracy of the method compared to visual analysis of structural OCT images. Alteration of myelinated fibers causes changes in the scattering properties of the white matter, which gets reflected in the nature of the received CP OCT signal. Visual assessment of structural CP OCT images and color-coded maps allows differentiating studied tissue types from each other, while usage of color-coded maps demonstrates higher diagnostic accuracy values in comparison with structural images (F-score = 0.85-0.86 and 0.81, respectively). Thus, the results of the study confirm the potential of using OCT as a neuronavigation tool during resections of brain tumors.

Melanoma and Nevi Subtype Histopathological Characterization with Optical Coherence Tomography.

Melanoma incidence has continued to rise in the latest decades, and the forecast is not optimistic. Non-invasive diagnostic imaging techniques such as optical coherence tomography (OCT) are largely studied; however, there is still no agreement on its use for the diagnosis of melanoma. For dermatologists, the differentiation of non-invasive (junctional nevus, compound nevus, intradermal nevus, and melanoma in-situ) versus invasive (superficial spreading melanoma and nodular melanoma) lesions is the key issue in their daily routine. This work performs a comparative analysis of OCT images using haematoxylin-eosin (HE) and anatomopathological features identified by a pathologist. Then, optical and textural properties are extracted from OCT images with the aim to identify subtle features that could potentially maximize the usefulness of the imaging technique in the identification of the lesion's potential invasiveness. Preliminary features reveal differences discriminating melanoma in-situ from superficial spreading melanoma and also between melanoma and nevus subtypes that pose a promising baseline for further research. Answering the final goal of diagnosing non-invasive versus invasive lesions with OCT does not seem feasible in the short term, but the obtained results demonstrate a step forward to achieve this.

Diagnosis of coronary layered plaque by deep learning

Healed coronary plaques, morphologically characterized by a layered phenotype, are signs of previous plaque destabilization and healing. Recent optical coherence tomography (OCT) studies demonstrated that layered plaque is associated with higher levels of local and systemic inflammation and rapid plaque progression. However, the diagnosis of layered plaque needs expertise in OCT image analysis and is susceptible to inter-observer variability. We developed a deep learning (DL) model for an accurate diagnosis of layered plaque. A Visual Transformer (ViT)-based DL model that integrates information from adjacent frames emulating the cardiologists who review consecutive OCT frames to make a diagnosis was developed and compared with the standard convolutional neural network (CNN) model. A total of 237,021 cross-sectional OCT images from 581 patients collected from 8 sites were used for training and internal validation, and 65,394 images from 292 patients collected from another site were used for external validation. In the five-fold cross-validation, the ViT-based model provided better performance (area under the curve [AUC]: 0.860; 95% confidence interval [CI]: 0.855–0.866) than the standard CNN-based model (AUC: 0.799; 95% CI: 0.792–0.805). The ViT-based model (AUC: 0.845; 95% CI: 0.837–0.853) also surpassed the standard CNN-based model (AUC: 0.791; 95% CI: 0.782–0.800) in the external validation. The ViT-based DL model can accurately diagnose a layered plaque, which could help risk stratification for cardiac events.

Self-training adversarial learning for cross-domain retinal OCT fluid segmentation

Accurate measurements of the size, shape and volume of macular edema can provide important biomarkers to jointly assess disease progression and treatment outcome. Although many deep learning-based segmentation algorithms have achieved remarkable success in semantic segmentation, these methods have difficulty obtaining satisfactory segmentation results in retinal optical coherence tomography (OCT) fluid segmentation tasks due to low contrast, blurred boundaries, and varied distribution. Moreover, directly applying a well-trained model on one device to test the images from other devices may cause the performance degradation in the joint analysis of multi-domain OCT images. In this paper, we propose a self-training adversarial learning framework for unsupervised domain adaptation in retinal OCT fluid segmentation tasks. Specifically, we develop an image style transfer module and a fine-grained feature transfer module to reduce discrepancies in the appearance and high-level features of images from different devices. Importantly, we transfer the target images to the appearance of source images to ensure that no image information of the source domain for supervised training is lost. To capture specific features of the target domain, we design a self-training module based on a discrepancy and similarity strategy to select the images with better segmentation results from the target domain and then introduce them into the source domain for the iterative training segmentation model. Extensive experiments on two challenging datasets demonstrate the effectiveness of our proposed method. In Particular, our proposed method achieves comparable results on cross-domain retinal OCT fluid segmentation compared with the state-of-the-art methods.