Detecting and quantifying high reflectance (HRF) in retinal patients

Abstract

The present application relates to a method for identifying high reflectance (HRF) in an eye of a patient, the method comprising: accessing one or more optical coherence tomography (OCT) scans of the retina of the eye of the patient; and inputting the one or more OCT scans into one or more machine learning models, the one or more machine learning models being trained to segment the one or more OCT scans to identify a set of highly reflective entities detectable from the one or more OCT scans. The method further includes determining one or more diameter measurements corresponding to each of the identified set of highly reflective entities based on the segmented one or more OCT scans; and identifying a high reflectance (HRF) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements satisfies a diameter threshold.

Description

[0001] Cross - Reference to Related Applications

[0002] This application claims priority to U.S. Provisional Patent Application No. 63 / 432,654, filed on Dec. 14, 2022, entitled "DETECTING AND QUANTIFYING HYPERREFLECTIVE FOCI (HRF) IN RETINAL PATIENTS", the entire content of which is hereby incorporated by reference herein. Technical Field

[0003] This application generally relates to hyperreflective foci (HRF), and more particularly to detecting and quantifying HRF in the retina of a patient's eye. Background Art

[0004] Hyperreflective foci (HRF) have been shown to be associated with various retinal diseases, such as diabetic retinopathy (DR), diabetic macular edema (DME), age - related macular degeneration (AMD), neovascular age - related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), and retinal vein occlusion (RVO). In fact, recent studies of HRF have demonstrated that these lesions may represent important biomarkers for retinal disease progression, prognosis, and treatment outcome. HRF can include discrete, well - demarcated lesions characterized by a reflectivity equal to or greater than that of the retinal pigment epithelium (RPE). HRF may be identified in some optical coherence tomography (OCT) scans. However, the quantification and accurate identification of HRF remain a significant challenge in advancing the understanding of the role of HRF in the pathogenesis of the above - mentioned retinal diseases.

[0005] Specifically, computationally-based segmentation of HRF remains a challenging and elusive task. For example, many existing computationally-based models struggle to distinguish HRF, such as from retinal blood vessels, hard exudates, and speckle noise, especially when in the presence of other retinal disease biomarkers. In fact, while computationally-based segmentation of HRF can ostensibly improve the ability to identify and study HRF, significant challenges remain in developing and deploying computationally-based segmentation models in clinical practice. Such challenges include a lack of sufficient training data, as manual annotation of HRF or other similar hyper-reflective substances can be time-consuming, costly, and prone to significant human error. Such challenges may further include a lack of a unified standard for defining and accurately identifying HRF, as many existing computationally-based models may identify any substance less than 100 micrometers (μm) (e.g., hard exudates, general hyper-reflective material (HRM), speckle noise) as HRF. Accordingly, techniques for accurately detecting and quantifying HRF in the retina of a patient's eye may be useful. Summary of the Invention

[0006] Embodiments of the present disclosure relate to one or more computing devices, methods, and non-transitory computer-readable media for detecting and quantifying hyper-reflective foci (HRF) in the retina of a patient's eye. In certain embodiments, one or more computing devices may access one or more optical coherence tomography (OCT) B-scans of the retina of the patient's eye. In certain embodiments, one or more computing devices may then input the one or more OCT B-scans into one or more machine learning models (e.g., semantic segmentation models) of an HRF segmentation and classification pipeline, wherein the one or more machine learning models (e.g., semantic segmentation models) may be trained to segment the one or more OCT B-scans to identify a set of hyper-reflective entities detectable from the one or more OCT B-scans. For example, in some embodiments, the one or more machine learning models (e.g., semantic segmentation models) may generate predictions of a segmentation map, wherein the segmentation map identifies the set of hyper-reflective entities. In some embodiments, the identified set of hyper-reflective entities may include various hyper-reflective entities, including, for example, hyper-reflective material (HRM), intraretinal hyper-reflective material (IHRM), and HRF.

[0007] In some embodiments, the one or more machine learning models (e.g., semantic segmentation model) can then output the set of identified highly reflective entities (e.g., HRM, IHRM, HRF) to a classification module (e.g., an algorithm based on image processing) of the HRF segmentation and classification pipeline. The classification module (e.g., an algorithm based on image processing) can then be utilized to determine one or more diameter measurements corresponding to each of the identified highly reflective entities in the set of identified highly reflective entities, and further identify the HRF in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a diameter threshold.

[0008] Specifically, according to embodiments of the present disclosure, the classification module (e.g., an algorithm based on image processing) can estimate the diameter of each of the identified highly reflective entities (e.g., HRM, IHRM, HRF) in the set, and classify each highly reflective entity having an estimated diameter of 50 micrometers (μm) or less as an HRF, and classify each highly reflective entity having an estimated diameter in the range of 50 μm to 100 μm as an IHRM. In some embodiments, the classification module can further classify each highly reflective entity having an estimated diameter greater than 100 μm as belonging to a third object category. In certain embodiments, based on mapping information of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid, the feature extraction module (e.g., an algorithm based on image processing) of the HRF segmentation and classification pipeline can then be utilized to calculate one or more volume measurements (e.g., volume, area, thickness, quantity, etc.) of the identified HRF within one or more corresponding ETDRS subfields.

[0009] In this way, the disclosed embodiments can provide an HRF segmentation and classification pipeline that can be suitable for accurately detecting and quantifying HRF in the retina of the eye of a patient, where the HRF is specifically defined as a highly reflective entity having an estimated diameter of 50 μm or less. In fact, by providing an HRF segmentation and classification pipeline suitable for accurately detecting and quantifying HRF in the retina of the eye of a patient, the present embodiments accurately and efficiently identify clinically significant biomarkers of visual acuity and morphological changes in many retinal diseases, such as diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), and retinal vein occlusion (RVO). Additionally, according to embodiments of the present disclosure, the provided HRF segmentation and classification pipeline can further be suitable for predicting retinal disease progression and patient treatment response using the detected and quantified HRF.

[0010] In some embodiments, the one or more computing devices may access one or more optical coherence tomography (OCT) scans of the retina of a patient's eye. In some embodiments, the one or more computing devices may input the one or more OCT scans into one or more machine learning models that are trained to segment the one or more OCT scans to identify a set of highly reflective entities detectable from the one or more OCT scans. In some embodiments, the set of highly reflective entities may include a collection of highly reflective material (HRM), intraretinal highly reflective material (IHRM), and HRF. In some embodiments, the one or more machine learning models may include at least one semantic segmentation model. In one embodiment, the at least one semantic segmentation model may include a U-Net architecture.

[0011] In some embodiments, the one or more computing devices may then determine one or more diameter measurements corresponding to each highly reflective entity in the identified set of highly reflective entities based on the segmented one or more OCT scans. In some embodiments, the one or more computing devices may then identify HRF in the retina of the patient's eye based on whether at least one of the one or more diameter measurements meets a diameter threshold. For example, in some embodiments, determining whether at least one of the one or more diameter measurements meets the diameter threshold may include: for each highly reflective entity in the identified set of highly reflective entities, associating an ellipse with the identified highly reflective entity, determining the diameter of the longest axis of the ellipse, and estimating at least one of the one or more diameter measurements based on the diameter of the longest axis of the ellipse.

[0012] In some embodiments, the one or more computing devices may identify HRFs in the retina of the patient's eye by identifying a subset of the identified set of highly reflective entities. For example, in one embodiment, the diameter threshold may include a minimum diameter of approximately 50 micrometers (μm). In some embodiments, the one or more computing devices may then identify intraretinal highly reflective material (IHRM) in the retina of the patient's eye based on whether at least one of the one or more diameter measurements meets a second diameter threshold. For example, in one embodiment, the second diameter threshold includes a diameter range of approximately 50 μm to 100 μm. In some embodiments, the classification module may further classify each highly reflective entity having an estimated diameter greater than 100 μm as belonging to a third object category. In some embodiments, identifying HRFs in the retina of the patient's eye may further include classifying the patient's eye as having at least one of: diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), or retinal vein occlusion (RVO).

[0013] In some embodiments, identifying HRFs in the retina of the patient's eye may further include accessing (ETDRS) grid mapping information that identifies one or more partitions of an Early Treatment Diabetic Retinopathy Study (ETDRS) grid, and at least partially determining one or more volume measurements of the identified HRFs based on the ETDRS grid mapping information. In some embodiments, the one or more volume measurements may include one or more of: the HRF volume in the retina of the patient's eye, the HRF area in the retina of the patient's eye, or the HRF thickness in the retina of the patient's eye.

[0014] For example, in some embodiments, the one or more computing devices may at least partially determine the HRF volume in the retina of the patient's eye relative to at least one of the identified one or more partitions based on the ETDRS grid mapping information. In one embodiment, determining the HRF volume in the retina of the patient's eye may include determining a decrease in the HRF volume in the retina of the patient's eye. In one embodiment, the at least one of the identified one or more partitions may include an outer retina partition of the ETDRS grid.

[0015] In some embodiments, the one or more computing devices may determine the number of HRFs in the retina of the patient's eye relative to at least one of the identified one or more partitions, at least in part based on the ETDRS grid mapping information. In some embodiments, the one or more computing devices may access a frontal image of the retina of the patient's eye, where the frontal image is associated with the one or more OCT scans. In some embodiments, the one or more computing devices may then map the identified HRFs to the frontal image, at least in part based on the ETDRS grid mapping information.

[0016] In some embodiments, the one or more computing devices may train the one or more machine learning models. For example, in some embodiments, training the one or more computing devices may include accessing an OCT scan data set of the retinas of the eyes of one or more patients. For example, in one embodiment, the OCT scan data set may include sparse annotations of HRFs in the retinas of the eyes of the one or more patients. In some embodiments, training the one or more computing devices may further include partitioning the OCT scan data set into a model training data set and a model validation data set, and training the one or more machine learning models based on the model training data set to segment OCT scans to identify a set of highly reflective entities detectable from the OCT scans. In one embodiment, the identified set of highly reflective entities may include highly reflective material (HRM).

[0017] In some embodiments, training the one or more computing devices may further include evaluating the one or more machine learning models based on the model validation data set. In some embodiments, training the one or more computing devices may further include identifying HRFs in the retinas of the eyes of the one or more patients based on whether one or more diameter measurements of the HRM meet a predetermined diameter threshold. In some embodiments, the sparse annotation of HRFs may include a boundary geometry enclosing multiple HRF instances, and thus training the one or more computing devices may further include adjusting the boundary geometry by reducing the size of the boundary geometry, where the size of the boundary geometry is reduced to annotate individual HRF instances.

[0018] In some embodiments, the one or more OCT scans can include one or more first OCT scans of the retina of the eye of the patient captured on an initial date. In one embodiment, the identified HRF can include a first HRF volume. In some embodiments, the one or more computing devices can access one or more second OCT scans of the retina of the eye of the patient. In some embodiments, the one or more computing devices can then input the one or more second OCT scans into the one or more machine learning models to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans. In some embodiments, the one or more computing devices can then determine one or more second diameter measurements corresponding to each highly reflective entity in the identified second set of highly reflective entities based on the segmented one or more second OCT scans.

[0019] In some embodiments, the one or more computing devices can then identify a second HRF volume in the retina of the eye of the patient based on whether at least one of the one or more second diameter measurements meets the diameter threshold. In one embodiment, the one or more computing devices can then determine whether the eye of the patient responds to treatment based on the second HRF volume. In another embodiment, the one or more computing devices can further determine the degree to which the eye of the patient responds to the treatment based on the second HRF volume. For example, in some embodiments, when the second HRF volume is less than the first HRF volume, the eye of the patient responds to the treatment. In some embodiments, the one or more second OCT scans are captured on one or more dates selected from the group consisting of: approximately 0.25 months, 0.5 months, 0.75 months, 1 month, 3 months, 6 months, 9 months, 12 months, 15 months, 18 months, 21 months, 24 months, 27 months, 30 months, 33 months, or 36 months from the initial date.

[0020] In some embodiments, the treatment can include an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof. In some embodiments, the anti-VEGF-A antibody can include faricimab-svoa. In some embodiments, the anti-Ang-2 antibody can include faricimab. In some embodiments, the anti-VEGF antibody can be selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium. In some embodiments, the one or more computing devices can identify an effective treatment regimen for an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof to treat the eye of the patient based on the volume or number of identified HRFs.

[0021] For example, in one embodiment, identifying the effective treatment regimen can include identifying a dose for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs. In another embodiment, identifying the effective treatment regimen can include identifying a schedule for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs. In another embodiment, identifying the effective treatment regimen can include identifying a duration for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] One or more of the figures included herein are in color in accordance with 37 CFR § 1.84. The color figures are necessary for illustrating the present invention.

[0023] Figure 1 A retinal segmentation, classification, and feature extraction network and system are shown in accordance with some embodiments disclosed herein.

[0024] Figure 2A A diagram of an inference stage of an HRF segmentation and classification pipeline suitable for detecting and quantifying HRFs in the retina of a patient's eye is shown in accordance with some embodiments disclosed herein.

[0025] Figure 2B A diagram of a training stage of an HRF segmentation and classification pipeline suitable for detecting and quantifying HRFs in the retina of a patient's eye is shown in accordance with some embodiments disclosed herein.

[0026] Figure 2C A diagram showing the classification stage of the HRF segmentation and classification pipeline according to some embodiments disclosed herein.

[0027] Figure 3A A flowchart showing a method for identifying high-reflectivity foci (HRF) in a patient's eye according to some embodiments disclosed herein.

[0028] Figure 3B A flowchart showing a method for determining whether a patient's eye responds to treatment based on changes in HRF volume or quantity according to some embodiments disclosed herein.

[0029] Figure 4A An enlarged example image of an OCT B-scan according to some embodiments disclosed herein.

[0030] Figure 4B An enlarged example image of a segmented and classified OCT B-scan according to some embodiments disclosed herein.

[0031] Figure 5A and 5B Respectively show enlarged example images of a first frontal image and a second frontal image according to some embodiments disclosed herein.

[0032] Figure 6 Shows an example computing system according to some embodiments disclosed herein.

[0033] Figure 7 Shows being included as Figure 6 A diagram of an example artificial intelligence (AI) architecture that is part of the example computing system. Detailed Description

[0034] Embodiments of the present disclosure relate to one or more computing devices, methods, and non-transitory computer-readable media for detecting and quantifying high-reflectivity foci (HRFs) in the retina of a patient's eye. In certain embodiments, one or more computing devices may access one or more optical coherence tomography (OCT) B-scans of the retina of a patient's eye. In certain embodiments, one or more computing devices may then input the one or more OCT B-scans into one or more machine learning models (e.g., semantic segmentation models) of an HRF segmentation and classification pipeline, wherein the one or more machine learning models (e.g., semantic segmentation models) may be trained to segment the one or more OCT B-scans to identify a set of high-reflectivity entities detectable from the one or more OCT B-scans. For example, in some embodiments, the one or more machine learning models (e.g., semantic segmentation models) may generate predictions of a segmentation map that identifies the set of high-reflectivity entities. In some embodiments, the identified set of high-reflectivity entities may include various high-reflectivity entities, including, for example, high-reflectivity material (HRM), intraretinal high-reflectivity material (IHRM), and HRF.

[0035] In certain embodiments, the one or more machine learning models (e.g., semantic segmentation models) may then output the identified set of high-reflectivity entities (e.g., HRM, IHRM, HRF) to a classification module (e.g., an image processing-based algorithm) of the HRF segmentation and classification pipeline. The classification module (e.g., an image processing-based algorithm) may then be utilized to determine one or more diameter measurements corresponding to each high-reflectivity entity in the identified set of high-reflectivity entities and further identify an HRF in the retina of the patient's eye based on whether at least one of the one or more diameter measurements meets a diameter threshold.

[0036] Specifically, according to embodiments of the present disclosure, the classification module (e.g., an algorithm based on image processing) can estimate the diameter of each of the identified highly reflective entities (e.g., HRM, IHRM, HRF) in the concentration, and classify each highly reflective entity having an estimated diameter of 50 micrometers (μm) or less as an HRF, and classify each highly reflective entity having an estimated diameter in the range of 50 μm to 100 μm as an IHRM. In some embodiments, the classification module can further classify each highly reflective entity having an estimated diameter greater than 100 μm as belonging to a third object category. In certain embodiments, then, based on the mapping information of the Early Treatment Diabetic Retinopathy Study (ETDRS) grid, the feature extraction module (e.g., an algorithm based on image processing) of the HRF segmentation and classification pipeline can be used to calculate one or more volume measurement results (e.g., volume, area, thickness, quantity, etc.) of the identified HRF within one or more corresponding ETDRS subfields.

[0037] In this way, the disclosed embodiments can provide an HRF segmentation and classification pipeline that can be suitable for accurately detecting and quantifying HRFs in the retina of a patient's eye, where an HRF is specifically defined as a highly reflective entity having an estimated diameter of 50 μm or less. In fact, by providing an HRF segmentation and classification pipeline suitable for accurately detecting and quantifying HRFs in the retina of a patient's eye, the present embodiments accurately and efficiently identify clinically significant biomarkers of visual acuity and morphological changes in many retinal diseases, such as diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), and retinal vein occlusion (RVO). Additionally, according to embodiments of the present disclosure, the provided HRF segmentation and classification pipeline can further be suitable for predicting retinal disease progression and patient treatment response using the detected and quantified HRFs.

[0038] Figure 1Disclosed is a retinal segmentation, classification, and feature extraction system 100 according to an embodiment of the present disclosure, which can be used to detect and quantify HRFs in the retina of a patient's eye. In certain embodiments, the retinal segmentation, classification, and feature extraction system 100 may include one or more retinal imaging platforms 102 and a retinal segmentation, classification, and feature extraction pipeline 104. For example, in some embodiments, one or more imaging platforms 102 may include one or more non-invasive retinal scan capture devices (e.g., ophthalmoscopes, scanning lasers, ultra-widefield fundus cameras, or other retinal scan capture modules), which may scan the patient's retina and generate one or more high-resolution 2D or 3D retinal scans 106.

[0039] For example, in some embodiments, the retinal scan 106 may include one or more OCT scans (e.g., time-domain OCT (TD-OCT) scans, spectral-domain OCT (SD-OCT) scans), which may be used to capture and present the depth of retinal layers. Specifically, in certain embodiments, when capturing an image of the patient's retina, one or more imaging platforms 102 may perform a series of one-dimensional (1D) scans (e.g., amplitude scans or "A scans") at different depths or positions and use the series of OCT A scans to generate a 2D cross-sectional image (e.g., luminance scan or "B scan") of the patient's three-dimensional (3D) retina. In certain embodiments, by closely and rapidly acquiring and generating OCT B scans, one or more imaging platforms 102 may further generate one or more volume images ("C scans") of the patient's three-dimensional (3D) retina.

[0040] In other embodiments, the retinal scan 106 can include one or more color fundus photography (CFP) images (e.g., multi-color 2D images of the retina, infrared (IR) 2D images of the retina), one or more retinal angiography scans (e.g., fluorescein angiography (FA) scans, OCT angiography (OCT-A) scans, ultra-wide field fluorescein angiography (UWFA) scans) images (e.g., ultra-wide field fluorescein angiography (UWFA) scans, indocyanine green angiography (ICGA) scans), one or more fundus autofluorescence (FAF) scans, one or more blue light autofluorescence (BAF) scans, and other similar retinal scans. In one embodiment, the retinal scan 106 (e.g., a number of OCT B-scans) can be captured by a retinal specialist (e.g., an ophthalmologist, an optometrist) during one or more patient visits to a clinical setting and one or more subsequent visits. In another embodiment, the retinal scan 106 (e.g., a number of OCT B-scans) can include a dataset of retinal scans (e.g., OCT B-scans) captured from various patients during one or more clinical trials. In one embodiment, the dataset of retinal scans (e.g., OCT B-scans) can be annotated and used to train one or more segmentation, classification, and feature extraction machine learning (ML) models or similar models.

[0041] In certain embodiments, as Figure 1 further depicted, one or more retinal imaging platforms 102 can then provide the retinal scan 106 (e.g., a number of OCT B-scans) to the retinal segmentation, classification, and feature extraction pipeline 104. In certain embodiments, the retinal segmentation, classification, and feature extraction pipeline 104 can include training modules and sub-modules 108, prediction modules and sub-modules 110, feature modules and sub-modules 112, and analysis modules and sub-modules 114 that can be utilized in one or more downstream uses 116. Although the training and execution of the training modules and sub-modules 108, prediction modules and sub-modules 110, feature modules and sub-modules 112, and analysis modules and sub-modules 114 can be discussed herein in a generally sequential manner (e.g., for purposes of brevity and illustration), it should be understood that the training modules and sub-modules 108, prediction modules and sub-modules 110, feature modules and sub-modules 112, and analysis modules and sub-modules 114 can be trained and / or executed according to an end-to-end deep learning process. For example, in certain embodiments, the training modules and sub-modules 108, prediction modules and sub-modules 110, feature modules and sub-modules 112, and analysis modules and sub-modules 114 can be trained and / or executed end-to-end (e.g., preprocessing, feature extraction and selection, optimization, prediction, decision-making, etc.) as a single integrated model for detecting and quantifying HRF in the retina of a patient's eye.

[0042] In view of the foregoing, in certain embodiments, during the training phase, a retinal scan 106 (e.g., a number of OCT B-scans) may be provided to the data processing functional block 118. In certain embodiments, the data processing functional block 118 may estimate or determine the quality of the retinal scan 106 (e.g., a number of OCT B-scans) and perform one or more transformations 120 and OCT B-scan extraction 122 based on the estimated quality to improve and / or enhance the quality of the retinal scan 106 (e.g., a number of OCT B-scans) or one or more features included in the retinal scan 106 (e.g., a number of OCT B-scans). For example, the retinal scan 106 may include a number of OCT B-scan extractions of the retinas of one or more patients, and the number of OCT B-scan extractions may be preprocessed to improve and / or enhance, for example, image quality, local contrast, image resolution, speckle noise, etc. Similarly, one or more transformations 120 may include one or more transformations (e.g., super-resolution algorithms, image alignment algorithms, pixel alignment algorithms, image stitching, etc.) to further improve and / or enhance the quality of the retinal scan 106 (e.g., a number of OCT B-scans).

[0043] In certain embodiments, as Figure 1 further depicted, the retinal segmentation, classification, and feature extraction pipeline 104 may include one or more artificial intelligence (AI) / machine learning (ML) accelerators 124A, 124B (e.g., one or more of a general-purpose processor, a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), a system-on-chip (SoC), a microcontroller, a field-programmable gate array (FPGA), a central processing unit (CPU), an application processor (AP), a vision processing unit (VPU), a neural processing unit (NPU), a neural decision processor (NDP), a deep learning processor (DLP), a tensor processing unit (TPU), a neuromorphic processing unit (NPU)), and the one or more artificial intelligence (AI) / machine learning (ML) accelerators may be suitable for hosting, executing, and / or processing various training modules and sub-modules 108, prediction modules and sub-modules 110, feature modules and sub-modules 112, and / or analysis and sub-modules 114.

[0044] In some embodiments, after transformation 120 and OCT B-scan extraction 122, as performed by data processing block 118, a retinal scan 106 (e.g., a number of OCT B-scans) may be provided to annotation processing block 126. For example, in some embodiments, a retinal scan 106 (e.g., a number of OCT B-scans) may include a data set of retinal scans (e.g., a number of OCT B-scans) that may be pre-annotated or sparsely annotated (e.g.) and used to train one or more deep learning models 130 for performing semantic segmentation (e.g., pixel-by-pixel segmentation) and classification to segment and annotate one or more layer features (e.g., layers of the retina), one or more fluid features, or one or more highly reflective entities detectable from the retinal scan 106 (e.g., a number of OCT B-scans).

[0045] For example, in certain embodiments, one or more deep learning models 130 may include a deep residual neural network (ResNet) image classification network (e.g., ResNet-34, ResNet-50, ResNet-101, ResNet-152), a full-resolution residual network (FRRN), a fully convolutional network (FCN) (e.g., U-Net), a pyramid scene parsing network (PSPNet), a fully convolutional dense neural network (FCDenseNet), a multi-path refinement network (RefineNet), a dilated convolutional network (e.g., DeepLabV3, DeepLabV+), a semantic segmentation network (SegNet), or other deep network convolutional network (DCNN) suitable for: performing semantic segmentation and classification to segment and annotate one or more layer features (e.g., layers of the retina), one or more fluid features, or highly reflective entities detectable from the retinal scan 106 (e.g., a number of OCT B-scans). In one embodiment, a performance monitoring block 132 may be provided to monitor and evaluate one or more deep learning models 130 during the training phase until the one or more deep learning models 130 are sufficiently trained.

[0046] In some embodiments, after one or more deep learning models 130 are fully trained to perform, for example, semantic segmentation (e.g., pixel-by-pixel segmentation) and classification to segment and annotate one or more layer features (e.g., layers of the retina), one or more fluid features, or one or more highly reflective entities detectable from a retinal scan 106 (e.g., a number of OCT B-scans), during the inference phase, a retinal scan 106 (e.g., a number of OCT B-scans) can be provided to a data processing functional block 134. The data processing functional block 134 can perform one or more transformations 136 and OCT B-scan extraction 138. For example, the retinal scan 106 can include a number of OCT B-scan extractions of the retinas of one or more patients, and the number of OCT B-scan extractions can be preprocessed to improve, for example, image quality, local contrast, image resolution, speckle noise, etc. Similarly, one or more transformations 136 can include one or more transformations (e.g., super-resolution algorithms, image alignment algorithms, pixel alignment algorithms, image stitching, etc.) to further improve the quality of the retinal scan 106 (e.g., a number of OCT B-scans).

[0047] In some embodiments, before providing the preprocessed retinal scan 106 (e.g., a number of OCT B-scans) to a prediction functional block 142, one or more fluid features or highly reflective entities in at least a subset of the retinal scan 106 (e.g., a number of B-scans) can be annotated by reading a central data input 140. In some embodiments, reading the central data input 140 can include annotations by one or more medical or scientific experts, such as from an ophthalmic reading center. For example, in one embodiment, the retinal scan 106 (e.g., a number of OCT B-scans) can be manually annotated by drawing boundary geometries or contours (e.g., representing fluid features or highly reflective entities).

[0048] In some embodiments, the layers of the retina and the pixel regions and corresponding labels of the fluid features or highly reflective entities can be predicted using one or more deep learning models 130. For example, one or more deep learning models 130 can generate predictions for one or more label maps 144 (e.g., including one or more OCT B-scans with segmented and annotated layers of the retina) and segmented fluid features or highly reflective entities 146 (e.g., fluid features and / or highly reflective entities segmented and annotated on one or more OCT B-scans).

[0049] In some embodiments, the layers of the retina can include one or more of the following: Bruch's membrane (BM), the border of the myoid and the inner segment of the ellipsoid (BMEIS), the ganglion cell layer - inner plexiform layer (GCL - IPL), the inner border outer photoreceptor (IB - OPR) layer, the outer border outer photoreceptor (OB - OPR) layer, the inner border retinal pigment epithelium (IB - RPE) layer, the outer border retinal pigment epithelium (OB - RPE) layer, the inner limiting membrane (ILM), the inner plexiform layer - inner nuclear layer (IPL - INL), the inner plexiform layer - outer nuclear layer (IPL - ONL), the inner segment / outer segment junction (ISJ - OSJ) layer, the outer plexiform layer - Henle fiber layer (OPL - HFL), or the retinal nerve fiber layer - ganglion cell layer (RNFL - GCL). Similarly, in some embodiments, the fluid features can include one or more of the following: intraretinal fluid (IRF), subretinal fluid (SRF), pigment epithelial detachment (PED), or subretinal hyperreflective material (SHRM). In certain embodiments, the hyperreflective entities can include hyperreflective material (HRM), intraretinal hyperreflective material (IHRM), or hyperreflective foci (HRF).

[0050] In certain embodiments, after generating predictions of one or more labeled maps 144 (e.g., including one or more OCT B - scans with segmented and annotated layers of the retina) and segmented fluid features or hyperreflective features 146 (e.g., IRF, SRF, PED, SHRM, IHRM, HRF segmented and labeled on one or more OCT B - scans), the one or more labeled maps 144 and the segmented fluid features or hyperreflective entities 146 can be provided to the feature calculation functional block 154. In certain embodiments, the feature calculation functional block 154 can be used to extract one or more volume measurements of layer features 156, fluid features 158, and hyperreflective entities 159.

[0051] For example, in some embodiments, one or more volume measurements of layer feature 156, fluid feature 158, and highly reflective entity 159 may include a matrix of volume, thickness, area, or quantity features for quantitatively measuring and analyzing layer feature 156 (e.g., BM, BMEIS, GCL-IPL, IB-OPR layer, OB-OPR layer, IB-RPE layer, OB-RPE layer, ILM, IPL-INL, IPL-ONL, ISJ-OSJ layer, OPL-HFL, RNFL-GCL), fluid feature 158 (e.g., IRF, SRF, PED, or SHRM), and highly reflective entity 159 (e.g., HRM, IHRM, HRF) of the retina based on nine macular subregions determined from the Early Treatment Diabetic Retinopathy Study (ETDRS) grid. In some embodiments, one or more volume measurements of layer feature 156, fluid feature 158, and highly reflective entity 159 may then be analyzed to determine any outliers 152 and longitudinal data dynamics 154.

[0052] In some embodiments, one or more volume measurements of layer feature 156, fluid feature 158, and highly reflective entity 159 may then be provided to machine learning (ML) functional block 160. In some embodiments, ML functional block 160 may be used to perform one or more dimensionality reduction, feature selection, and classification tasks based on one or more volume measurements of layer feature 156, fluid feature 158, and highly reflective entity 159, and retina segmentation, classification, and feature extraction pipeline 104 may generate one or more final outputs.

[0053] For example, according to embodiments of the present disclosure, retina segmentation, classification, and feature extraction pipeline 104 may generate one or more of the following final outputs based on one or more volume measurements of layer feature 156, fluid feature 158, and highly reflective entity 159: classifying a patient as having one or more of diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), and retinal vein occlusion (RVO). In some embodiments, retina segmentation, classification, and feature extraction pipeline 104 may further generate one or more of the following final outputs: classifying one or more treatments for treating the retina of an eye of a patient with a retinal disease, determining the risk of progression of a retinal disease in an eye of a patient, or identifying an effective treatment regimen for treating the retina of an eye of a patient with a retinal disease.

[0054] In some embodiments, one or more final outputs can then be provided for downstream use, such as by clinician 170A (e.g., ophthalmologist, optometrist), biomarker scientist 170B, data scientist 170C, or for data storage and / or sharing 170D. For example, in some embodiments, a report can be generated based on one or more final outputs of the retinal segmentation, classification, and feature extraction pipeline 104. For example, in one embodiment, the report can include a clinical report that can be associated with one or more retinal patients, and the clinical report will be provided and displayed, for example, to clinician 170A (e.g., ophthalmologist, optometrist) for the study and / or diagnosis, prognosis, and treatment of one or more retinal patients. In another embodiment, the report can include an interpretability and / or explainability report that can be associated with the retinal segmentation, classification, and feature extraction pipeline 104, and the interpretability and / or explainability report will be provided and displayed, for example, to one or more data scientists 170C for determining and elucidating the prediction and decision-making behavior of the retinal segmentation, classification, and feature extraction pipeline 104.

[0055] Figure 2A FIG. 200A shows an inference phase of an HRF segmentation and classification pipeline suitable for detecting and quantifying HRF in the retina of a patient's eye, according to an embodiment of the present disclosure. In one embodiment, the HRF segmentation and classification pipeline can be a subset of the retinal segmentation, classification, and feature extraction pipeline 104 as discussed above with respect to Figure 1 As Figure 2A described, in certain embodiments, one or more OCT scans 202A of the retina of a patient's eye can be accessed and input into a segmentation machine learning model 204A. For example, in some embodiments, one or more OCT scans 202A can each include, for example, an OCT B-scan of the retina of a patient's eye. In certain embodiments, the segmentation model 204A can include a semantic segmentation model (e.g., FRRN, FCN, U-Net, PSPNet, FCDenseNet, RefineNet, DeepLabV3, DeepLabV+, SegNet, or a similar semantic segmentation DCNN), which can be trained to generate one or more predicted segmentation maps that identify a set of hyper-reflective entities. For example, in some embodiments, the identified set of hyper-reflective entities can include HRM, IHRM, HRF, or other similar HRMs that may be associated with one or more retinal diseases.

[0056] In some embodiments, the set of identified highly reflective entities can then be input into classification module 206A. For example, in some embodiments, classification module 206A can include an image processing-based algorithm or similar process that can be used to perform diameter measurements on the highly reflective entities (e.g., HRM, IHRM, HRF) identified by segmentation model 204A. For example, as will be further described with respect to Figure 2C the diameter measurement can be performed by classification module 206A by associating an ellipse with each identified highly reflective entity (e.g., HRM, IHRM, HRF) and determining the diameter of the longest axis of the ellipse associated with the identified highly reflective entity (e.g., HRM, IHRM, HRF).

[0057] In some embodiments, classification module 206A can then estimate the diameter of each identified highly reflective entity (e.g., HRM, IHRM, HRF) based on the determined diameter of the corresponding ellipse associated with each identified highly reflective entity. In some embodiments, classification module 206A can then identify (e.g., classify) HRF 208 in the retina of the patient's eye based on whether the estimated diameter (e.g., estimated based on the determined diameter of the corresponding ellipse associated with each identified highly reflective entity) meets a diameter threshold. In one embodiment, the diameter threshold can be such a diameter or range of diameters that is defined to distinguish HRF from other highly reflective entities (e.g., HRM, IHRM, or other HRM) that may be present in one or more OCT scans 202A and segmented by segmentation model 204A.

[0058] For example, in some embodiments, when the estimated diameter of the highly reflective entity includes a diameter of about 50 μm or less, classification module 206A can classify the highly reflective entity as HRF 208. In another embodiment, when the estimated diameter of the highly reflective entity includes a diameter of about 40 μm or less, about 35 μm or less, about 30 μm or less, about 25 μm or less, about 20 μm or less, about 15 μm or less, or about 10 μm or less, classification module 206A can classify the highly reflective entity as HRF 208. In other embodiments, when the estimated diameter of the highly reflective entity includes a diameter in the range of about 50 μm to 100 μm, classification module 206A can classify the highly reflective entity as IHRM. In another embodiment, when the estimated diameter of the highly reflective entity includes a diameter in the range of about 50 μm to 90 μm, about 50 μm to 80 μm, about 50 μm to 70 μm, or about 50 μm to 60 μm, classification module 206A can classify the highly reflective entity as IHRM.

[0059] In some embodiments, to further refine the highly reflective entity identified as HRF 208, predictions of one or more labeled maps 144 (e.g., one or more OCT B-scans including segmented and annotated layers of the retina) can be utilized to narrow the region within the segmented OCT B-scan where the highly reflective entity identified as HRF 208 can be identified. For example, in some embodiments, highly reflective entities identified as HRF 208 detected outside the boundaries of one or more specific segmented retinal layers can be discarded, and only highly reflective entities identified as HRF 208 within the boundaries of one or more specific segmented retinal layers can be passed to the feature extraction module 210A. For example, in one embodiment, only highly reflective entities identified as HRF 208 detected between the ILM layer and the OPL-HFL layer can correspond to the identified HRF 208 in the inner retina, and only highly reflective entities identified as HRF 208 detected between the OPL-HFL layer and the RPE layer can correspond to the identified HRF 208 in the outer retina.

[0060] In some embodiments, the identified HRF 208 can then be provided to the feature extraction module 210A. For example, in some embodiments, the feature extraction module 210A can include an image processing-based algorithm that can be used to determine one or more volume measurements 212 of the identified HRF 208. For example, in certain embodiments, based on the ETDRS grid and the mapping information of the identified HRF 208, the feature extraction module 210A (e.g., the image processing-based algorithm) can calculate one or more volume measurements 212, including, for example, the total volume of the identified HRF 208, the volume of the identified HRF 208 relative to one or more partitions of the ETDRS grid, the total area of the identified HRF 208, the area of the identified HRF 208 relative to one or more partitions of the ETDRS grid, the thickness of the identified HRF 208 relative to one or more partitions of the ETDRS grid, the total number of the identified HRF 208, the number of the identified HRF 208 relative to one or more partitions of the ETDRS grid, etc.

[0061] In this manner, embodiments of the present disclosure can provide an HRF segmentation and classification pipeline that can be suitable for accurately detecting and quantifying HRF 208 in the retina of a patient's eye, where HRF 208 can be specifically defined as a highly reflective entity having an estimated diameter of 50 μm or less. In fact, by providing an HRF segmentation and classification pipeline suitable for accurately detecting and quantifying HRF 208 in the retina of a patient's eye, the present embodiment accurately and efficiently identifies clinically significant biomarkers of visual acuity and morphological changes in many retinal diseases such as DR, DME, AMD, nAMD, GA, MA, and RVO. Additionally, according to embodiments of the present disclosure, the provided HRF segmentation and classification pipeline can further be suitable for predicting retinal disease progression and patient treatment response using the detected and quantified HRF 208.

[0062] Figure 2B FIG. 200B shows a training phase of an HRF segmentation and classification pipeline suitable for detecting and quantifying HRF in the retina of a patient's eye according to an embodiment of the present disclosure. As Figure 2B depicted, in certain embodiments, the HRF segmentation and classification pipeline can include a preprocessing module 214 and a segmentation model 204B. The segmentation model 204B can correspond to the segmentation model 204A discussed above with respect to Figure 2A In some embodiments, the segmentation model 204B can include one or more machine learning models that can be trained end-to-end to identify HRM.

[0063] As Figure 2B depicted, in certain embodiments, a dataset of sparsely annotated OCT scans 202B of the retinas of one or more patients' eyes can be accessed and preprocessed by the preprocessing module 214. For example, in some embodiments, the dataset of sparsely annotated OCT scans 202B can include an OCT B-scan dataset in which highly reflective entities have been sparsely annotated by one or more human graders for training the segmentation model 204B. In one embodiment, the sparse annotation of the highly reflective entities can include a boundary geometry that encloses a certain number of highly reflective entities. In certain embodiments, the preprocessing module 214 can then preprocess the dataset of sparsely annotated OCT scans 202B by: adjusting the boundary geometry to reduce the size of the boundary geometry so as to annotate only a single highly reflective entity (e.g., as opposed to a boundary geometry that encloses multiple highly reflective entities simultaneously).

[0064] For example, in some embodiments, the preprocessing module 214 may utilize an Otsu thresholding algorithm (e.g., dynamic thresholding) or other similar image processing-based algorithms that may be suitable for binarizing a dataset of sparsely annotated OCT scans 202B based on pixel intensity. The boundary geometry may then be shrunk to enclose the highly reflective entity with the brightest intensity, which may better represent the HRF or IHRM for training the segmentation model 204B. In certain embodiments, the segmentation model 204B may then be trained using the dataset of sparsely annotated OCT scans 202B preprocessed by the preprocessing module 214. For example, in certain embodiments, the preprocessed dataset of sparsely annotated OCT scans 202B may be partitioned into a model training dataset and a model validation dataset. In certain embodiments, the model training dataset may be input into the segmentation model 204B for training the segmentation model 204B to identify a set of highly reflective entities 216 (e.g., HRM, IHRM, HRF). In certain embodiments, once trained, the model validation dataset may then be utilized to evaluate the performance of the segmentation model 204B.

[0065] Figure 2C FIG. 200C shows the classification stage of the HRF segmentation and classification pipeline according to embodiments of the present disclosure. In certain embodiments, as previously discussed with respect to Figure 2A the classification stage may be performed by the classification module 206B. For example, as Figure 2C depicted, after the highly reflective entities 218A, 218B, and 218C (e.g., HRM, IHRM, HRF) are identified by the segmentation model 204A, for example, one or more diameter measurements may be performed on the highly reflective entities 218A, 218B, and 218C (e.g., HRM, IHRM, HRF). In certain embodiments, the one or more diameter measurements may be performed by associating an ellipse 220 with each of the identified highly reflective entities 218A, 218B, and 218C (e.g., HRM, IHRM, HRF) and determining the diameter of the longest axis 222 of the ellipse 220 associated with each of the highly reflective entities 218A, 218B, and 218C (e.g., HRM, IHRM, HRF). Specifically, in some embodiments, for example, the HRF instance may be assumed to be an ellipse having a length α measured along the longest axis 222 of the ellipse 220.

[0066] In some embodiments, the HRFs in the retina of a patient's eye can then be identified (e.g., classified) based on whether one or more diameter measurements meet a diameter threshold. According to embodiments of the present disclosure, when one or more diameter measurements of a hyper-reflective entity include a diameter of 50 μm or less (e.g., α = ≤50 μm), the hyper-reflective entity is identified (e.g., classified) as an HRF. In other embodiments, when one or more diameter measurements of a hyper-reflective entity include a diameter in the range of approximately 50 μm to 100 μm (e.g., 50 μm < α ≤ 100 μm), the hyper-reflective entity is identified (e.g., classified) as an IHRM. In one embodiment, as Figure 2C further depicted, the volume of an identified HRF instance can be calculated as the product of the actual 2D area and the depth , where the depth can be defined as the distance between OCT B-scans.

[0067] Figure 3A FIG. shows a flowchart of a method 300A for detecting and quantifying HRFs in the retina of a patient's eye according to the disclosed embodiments. Method 300A can be carried out using one or more processing devices (e.g., the (one or more) computing devices and artificial intelligence architectures discussed below with respect to Figure 6 and 7 ), which can include hardware (e.g., a general-purpose processor, a graphics processing unit (GPU), an application-specific integrated circuit (ASIC), a system-on-chip (SoC), a microcontroller, a field-programmable gate array (FPGA), a central processing unit (CPU), an application processor (AP), a vision processing unit (VPU), a neural processing unit (NPU), a neural decision processor (NDP), a deep learning processor (DLP), a tensor processing unit (TPU), a neuromorphic processing unit (NPU), or any other artificial intelligence (AI) / machine learning (ML) accelerator device suitable for processing medical data and making one or more predictions or decisions based thereon), firmware (e.g., microcode), or some combination thereof.

[0068] Method 300A may include, at block 302, one or more processing devices accessing one or more optical coherence tomography (OCT) scans of a retina of a patient's eye. For example, one or more processing devices may receive one or more OCT B-scans of a retina of a patient's eye. Method 300A may include, at block 304, one or more processing devices inputting the one or more OCT scans into one or more machine learning models that are trained to segment the one or more OCT scans to identify a set of highly reflective entities detectable from the one or more OCT scans. For example, in some embodiments, one or more OCT B-scans of a retina of a patient's eye may be input into a semantic segmentation model (e.g., FRRN, FCN, U-Net, PSPNet, FCDenseNet, RefineNet, DeepLabV3, DeepLabV+, SegNet, or a similar semantic segmentation DCNN), which may generate one or more segmentation maps identifying a set of highly reflective entities (e.g., HRM, IHRM, HRF).

[0069] Method 300A may include, at block 306, one or more processing devices determining one or more diameter measurements corresponding to each highly reflective entity in the identified set of highly reflective entities based on the segmented one or more OCT scans. For example, in some embodiments, one or more processing devices may perform diameter measurements on highly reflective entities (e.g., HRM, IHRM, HRF) identified by the semantic segmentation model. For example, the diameter measurement may be performed by associating an ellipse with each identified highly reflective entity (e.g., HRM, IHRM, HRF) and determining the diameter of the longest axis of the ellipse associated with the identified highly reflective entity (e.g., HRM, IHRM, HRF).

[0070] Method 300A may include, at block 308, one or more processing devices identifying a highly reflective focus (HRF) in the retina of the patient's eye based on whether at least one of the one or more diameter measurements meets a diameter threshold. For example, in some embodiments, when one or more diameter measurements of a highly reflective entity include a minimum diameter of approximately 50 μm, one or more processing devices may classify the highly reflective entity as an HRF. In other embodiments, when one or more diameter measurements of a highly reflective entity include a diameter in the range of approximately 50 μm to 100 μm, one or more processing devices may classify the highly reflective entity as an IHRM. In certain embodiments, when one or more diameter measurements of a highly reflective entity include a diameter greater than 100 μm, one or more processing devices may discard any highly reflective entity.

[0071] Figure 3BFIG. 300B is a flowchart of a method for determining whether a patient's eye responds to treatment based on changes in HRF volume or quantity according to the disclosed embodiments. Method 300B may utilize one or more processing devices (e.g., one or more computing devices and artificial intelligence architectures discussed below with respect to Figure 6 and 7 ), which may include hardware (e.g., general-purpose processors, graphics processing units (GPUs), application-specific integrated circuits (ASICs), systems-on-a-chip (SoCs), microcontrollers, field-programmable gate arrays (FPGAs), central processing units (CPUs), application processors (APs), vision processing units (VPUs), neural processing units (NPUs), neural decision processors (NDPs), deep learning processors (DLPs), tensor processing units (TPUs), neuromorphic processing units (NPUs), or any other artificial intelligence (AI) / machine learning (ML) accelerator device suitable for processing medical data and making one or more predictions or decisions based thereon), firmware (e.g., microcode), or some combination thereof.

[0072] Method 300B may include, at block 310, one or more processing devices receiving one or more second optical coherence tomography (OCT) scans of the retina of the patient's eye. For example, one or more processing devices may receive one or more second OCT B-scans of the retina of the patient's eye, e.g., corresponding to OCT B-scans captured after a certain period of time during which the patient has received one or more treatments and / or treatment regimens. Method 300B may include, at block 312, one or more processing devices inputting the one or more second OCT scans into one or more machine learning models trained to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans. For example, in some embodiments, one or more second OCT B-scans of the retina of the patient's eye may be input into a semantic segmentation model (e.g., FRRN, FCN, U-Net, PSPNet, FCDenseNet, RefineNet, DeepLabV3, DeepLabV+, SegNet, or a similar semantic segmentation DCNN), which may generate one or more segmentation maps identifying a set of highly reflective entities (e.g., HRM, IHRM, HRF).

[0073] Method 300B may include, at block 314, one or more processing devices determining one or more second diameter measurements corresponding to each of the highly reflective entities in the identified second set of highly reflective entities based on the segmented one or more second OCT scans. For example, in some embodiments, one or more processing devices may perform diameter measurements on highly reflective entities (e.g., HRM, IHRM, HRF) identified by a semantic segmentation model. For example, the diameter measurement may be performed by associating an ellipse with each identified highly reflective entity (e.g., HRM, IHRM, HRF) and determining the diameter of the longest axis of the ellipse associated with the identified highly reflective entity (e.g., HRM, IHRM, HRF).

[0074] Method 300B may include, at block 316, one or more processing devices identifying a second highly reflective focus (HRF) volume or number in the retina of the patient's eye based on whether at least one of the one or more second diameter measurements meets a diameter threshold. For example, in some embodiments, one or more processing devices may classify a highly reflective entity as an HRF when one or more diameter measurements of the highly reflective entity include a minimum diameter of approximately 50 μm. In other embodiments, one or more processing devices may classify a highly reflective entity as an IHRM when one or more diameter measurements of the highly reflective entity include a diameter in the range of approximately 50 μm to 100 μm.

[0075] Method 300B may include, at block 318, one or more processing devices determining whether the patient's eye is responsive to treatment based on the second HRF volume or number. For example, in some embodiments, the second HRF volume or number detected from one or more second OCT B-scans may be compared to the first HRF volume or number detected from one or more first OCT B-scans of the retina of the patient's eye captured before the patient has received one or more treatments and / or treatment regimens. In certain embodiments, a decrease in the HRF volume or number may indicate that one or more treatments and / or treatment regimens are effective.

[0076] Figure 4A An enlarged example image of an OCT B-scan 400A in accordance with an embodiment of the present disclosure is shown. In certain embodiments, OCT B-scan 400A may be an OCT B-scan of the retina of a patient's eye, where the retina of the patient's eye includes an undetected and unquantified HRF in accordance with an embodiment of the present disclosure.

[0077] Figure 4BDisplays an enlarged exemplary image of a segmented and classified OCT B-scan 400B according to an embodiment of the present disclosure. In some embodiments, the segmented OCT B-scan 400B can be a segmented and classified OCT B-scan of the retina of a patient's eye, where the segmented and classified OCT B-scan 400B shows the results of the segmentation and classification of the OCT B-scan 400A discussed above with respect to Figure 4A As depicted, according to an embodiment of the present disclosure, the segmented and classified OCT B-scan 400B identifies the HRF (e.g., as shown by the yellow objects that appear rectangular and / or square).

[0078] Figure 5A and 5B respectively display enlarged exemplary images of a front image 500A and a front image 500B according to an embodiment of the present disclosure. In some embodiments, the front image 500A and the front image 500B can each include a rendering of a front image generated after segmentation, classification, and feature extraction of one or more OCT B-scans as generally described herein. As depicted, compared to the volume and number of HRF (e.g., as shown by the yellow objects or lighter-colored objects) and IHRM (e.g., as shown by the blue objects or darker-colored objects) depicted in the front image 500B, the front image 500A can include a greater volume and number of HRF (e.g., as shown by the yellow objects or lighter-colored objects) and IHRM (e.g., as shown by the blue objects or darker-colored objects). In one embodiment, the front image 500A can include an exemplary front image of a patient's eye before any treatment and / or treatment regimen has been received, while the front image 500B can include an exemplary front image of a patient's eye after treatment and / or treatment regimen has been received.

[0079] In one embodiment, the treatment can include an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or some combination thereof. For example, in one embodiment, the anti-VEGF-A antibody can include faricimab, and the anti-Ang-2 antibody can include faricimab. In another embodiment, the anti-VEGF antibody can be selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium.

[0080] Figure 6Illustrates an example of one or more computing devices 600 according to the disclosed embodiments, which may be used to detect and quantify HRFs in the retina of a patient's eye. In certain embodiments, one or more computing devices 600 may perform one or more steps of one or more of the methods described or illustrated herein. In certain embodiments, one or more computing devices 600 provide the functionality described or illustrated herein. In certain embodiments, software running on one or more computing devices 600 performs one or more steps of one or more of the methods described or illustrated herein, or provides the functionality described or illustrated herein. Certain embodiments include one or more portions of one or more computing devices 600.

[0081] The present disclosure contemplates any suitable number of computing systems 600. The present disclosure contemplates one or more computing devices 600 in any suitable physical form. By way of example and not limitation, one or more computing devices 600 may be an embedded computing system, a system on a chip (SOC), a single board computer system (SBC) (e.g., a computer on module (COM) or a system on module (SOM)), a desktop computer system, a laptop or notebook computer system, an interactive kiosk, a mainframe, a computer system network, a mobile phone, a personal digital assistant (PDA), a server, a tablet computer system, an augmented / virtual reality device, or a combination of two or more thereof. In appropriate instances, one or more computing devices 600 may: be integrated or distributed; span multiple locations; span multiple machines; span multiple data centers; or reside in a cloud, which may include one or more cloud components in one or more networks.

[0082] In appropriate instances, one or more computing devices 600 may perform one or more steps of one or more of the methods described or illustrated herein without substantial spatial or temporal limitations. By way of example and not limitation, one or more computing devices 600 may perform one or more steps of one or more of the methods described or illustrated herein in real time or in batch mode. In appropriate instances, one or more computing devices 600 may perform one or more steps of one or more of the methods described or illustrated herein at different times or in different locations.

[0083] In some embodiments, one or more computing devices 600 include a processor 602, a memory 604, a database 606, an input / output (I / O) interface 608, a communication interface 610, and a bus 612. Although this disclosure describes and shows a particular computer system having a particular number of particular components in a particular arrangement, this disclosure contemplates any suitable computer system having any suitable number of any suitable components in any suitable arrangement. In some embodiments, the processor 602 includes hardware for executing instructions, such as those that make up a computer program. By way of example and not limitation, to execute instructions, the processor 602 may: retrieve (or fetch) instructions from an internal register, an internal cache, the memory 604, or the database 606; decode and execute those instructions; and then write one or more results to an internal register, an internal cache, the memory 604, or the database 606. In some embodiments, the processor 602 may include one or more internal caches for data, instructions, or addresses. In appropriate instances, this disclosure contemplates a processor 602 that includes any suitable number of any suitable internal caches. By way of illustration and not limitation, the processor 602 may include one or more instruction caches, one or more data caches, and one or more translation lookaside buffers (TLBs). The instructions in the instruction cache may be copies of the instructions in the memory 604 or the database 606, and the instruction cache may accelerate the retrieval of these instructions by the processor 602.

[0084] The data in the data cache may be: a copy of the data in the memory 604 or the database 606 for the instructions being executed at the processor 602 to operate on; the result of a previous instruction executed at the processor 602 for a subsequent instruction executed at the processor 602 to access or write to the memory 604 or the database 606; or other suitable data. The data cache may accelerate the read or write operations of the processor 602. The TLB may accelerate the virtual address translation of the processor 602. In some embodiments, the processor 602 may include one or more internal registers for data, instructions, or addresses. In appropriate instances, this disclosure contemplates a processor 602 that includes any suitable number of any suitable internal registers. In appropriate instances, the processor 602 may include one or more arithmetic logic units (ALUs); may be a multi-core processor; or may include one or more processors 602. Although this disclosure describes and shows a particular processor, this disclosure contemplates any suitable processor.

[0085] In some embodiments, the memory 604 includes a main memory that stores instructions for execution by the processor 602 or data for the processor 602 to operate on. By way of example and not limitation, one or more computing devices 600 may load instructions from a database 606 or another source (such as, for example, another one or more computing devices 600) into the memory 604. Then, the processor 602 may load instructions from the memory 604 into internal registers or an internal cache. To execute the instructions, the processor 602 may retrieve the instructions from the internal registers or the internal cache and decode the instructions. During or after the execution of the instructions, the processor 602 may write one or more results (which may be intermediate results or final results) to the internal registers or the internal cache. Then, the processor 602 may write one or more of those results to the memory 604.

[0086] In some embodiments, the processor 602 executes instructions only in one or more internal registers, the internal cache, or the memory 604 (and not in the database 606 or elsewhere) and operates on data only in one or more internal registers, the internal cache, or the memory 604 (and not in the database 606 or elsewhere). One or more memory buses (each of which may include an address bus and a data bus) may couple the processor 602 to the memory 604. The bus 612 may include one or more memory buses, as described below. In some embodiments, one or more memory management units (MMUs) reside between the processor 602 and the memory 604 and facilitate access to the memory 604 requested by the processor 602. In some embodiments, the memory 604 includes random access memory (RAM). Where appropriate, the RAM may be volatile memory. Where appropriate, the RAM may be dynamic RAM (DRAM) or static RAM (SRAM). Additionally, where appropriate, the RAM may be single-port or multi-port RAM. The present disclosure contemplates any suitable RAM. Where appropriate, the memory 604 may include one or more memory devices 604. Although the present disclosure describes and illustrates particular memories, the present disclosure contemplates any suitable memory.

[0087] In some embodiments, database 606 includes a mass storage device for data or instructions. By way of example and not limitation, database 606 can include a hard disk drive (HDD), a floppy disk drive, flash memory, an optical disc, a magneto-optical disc, magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of them. In appropriate cases, database 606 can include removable or non-removable (or fixed) media. In appropriate cases, database 606 can be internal or external to one or more computing devices 600. In some embodiments, database 606 is a non-volatile solid-state memory. In some embodiments, database 606 includes a read-only memory (ROM). In appropriate cases, the ROM can be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable PROM (EEPROM), an electrically rewritable ROM (EAROM), flash memory, or a combination of two or more of them. The present disclosure contemplates a mass database 606 in any suitable physical form. In appropriate cases, database 606 can include one or more storage control units that facilitate communication between processor 602 and database 606. In appropriate cases, database 606 can include one or more databases 606. Although the present disclosure describes and illustrates particular storage devices, the present disclosure contemplates any suitable storage device.

[0088] In some embodiments, I / O interface 608 includes hardware, software, or both that provide one or more interfaces for communication between one or more computing devices 600 and one or more I / O devices. In appropriate cases, one or more of the computing devices 600 can include one or more of these I / O devices. One or more of these I / O devices can enable communication between a person and one or more computing devices 600. By way of example and not limitation, I / O devices can include a keyboard, a keypad, a microphone, a monitor, a mouse, a printer, a scanner, a speaker, a still camera, a stylus, a tablet computer, a touch screen, a trackball, a video camera, another suitable I / O device, or a combination of two or more of them. The I / O devices can include one or more sensors. The present disclosure contemplates any suitable I / O devices and any suitable I / O interface 608 therefor. In appropriate cases, I / O interface 608 can include one or more devices or software drivers that enable processor 602 to drive one or more of these I / O devices. In appropriate cases, I / O interface 608 can include one or more I / O interfaces 608. Although the present disclosure describes and illustrates particular I / O interfaces, the present disclosure encompasses any suitable I / O interface.

[0089] In some embodiments, communication interface 610 includes hardware, software, or both that provide one or more interfaces for communication (such as, for example, packet-based communication) between one or more computing devices 600 and one or more other computing devices 600 or one or more networks. By way of example and not limitation, communication interface 610 may include a network interface controller (NIC) or network adapter for communicating with an Ethernet or other wire-based network, or a wireless NIC (WNIC) or wireless adapter for communicating with a wireless network (such as a WI-FI network). The present disclosure contemplates any suitable network and any suitable communication interface 610 therefor.

[0090] By way of example and not limitation, one or more computing devices 600 may communicate with an ad hoc network, a personal area network (PAN), a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), one or more portions of the Internet, or a combination of two or more thereof. One or more portions of one or more of these networks may be wired or wireless. By way of example, one or more computing devices 600 may communicate with a wireless PAN (WPAN) (such as, for example, a BLUETOOTH WPAN), a WI-FI network, a WI-MAX network, a cellular telephone network (such as, for example, a Global System for Mobile Communications (GSM) network), other suitable wireless networks, or a combination of two or more thereof. In appropriate instances, one or more computing devices 600 may include any suitable communication interface 610 for any of these networks. In appropriate instances, communication interface 610 may include one or more communication interfaces 610. Although the present disclosure describes and illustrates particular communication interfaces, the present disclosure contemplates any suitable communication interface.

[0091] In some embodiments, bus 612 includes hardware, software, or both that couple components of one or more computing devices 600 to each other. By way of example and not limitation, bus 612 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an INFINIBAND interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCIe) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, another suitable bus, or a combination of two or more thereof. In appropriate instances, bus 612 may include one or more buses 612. Although the present disclosure describes and illustrates particular buses, the present disclosure contemplates any suitable bus.

[0092] In this document, one or more computer-readable non-transitory storage media may include one or more semiconductor-based or other integrated circuits (ICs) (such as, for example, field-programmable gate arrays (FPGAs) or application-specific ICs (ASICs)), hard disk drives (HDDs), hybrid hard disk drives (HHDs), optical discs, optical disc drives (ODDs), magneto-optical discs, magneto-optical disc drives, floppy disks, floppy disk drives (FDDs), magnetic tapes, solid-state drives (SSDs), RAM drives, Secure Digital cards or drives, any other suitable computer-readable non-transitory storage media, or any suitable combination of two or more of them. In appropriate cases, the computer-readable non-transitory storage media may be volatile storage media, non-volatile storage media, or a combination of volatile storage media and non-volatile storage media.

[0093] Figure 7 FIG. 700 shows an example artificial intelligence (AI) architecture 702 according to an embodiment of the present disclosure (which may be included as part of one or more computing devices 600 as discussed above with respect to Figure 6 ), and this example artificial intelligence (AI) architecture can be used to detect and quantify HRFs in the retina of a patient's eye. In some embodiments, the AI architecture 702 may be implemented using, for example, one or more processing devices, which may include hardware (e.g., general-purpose processors, graphics processing units (GPUs), application-specific integrated circuits (ASICs), system-on-chips (SoCs), microcontrollers, field-programmable gate arrays (FPGAs), central processing units (CPUs), application processors (APs), vision processing units (VPUs), neural processing units (NPUs), neural decision processors (NDPs), deep learning processors (DLPs), tensor processing units (TPUs), neuromorphic processing units (NPUs), and / or other artificial intelligence (AI) / machine learning (ML) accelerator devices suitable for processing various data and making one or more predictions or decisions based thereon), software (e.g., instructions running / executing on one or more processing devices), firmware (e.g., microcode), or some combination thereof.

[0094] In some embodiments, as shown by Figure 7Depicted, the AI architecture 702 can include a machine learning (ML) model 704, a natural language processing (NLP) model 706, an expert system 708, a computer-based vision model 710, speech recognition models and functions 712, a planning model 714, and robotic models and functions 716. In certain embodiments, the ML model 704 can include any statistical-based models that can be suitable for finding patterns in large amounts of data (e.g., “big data” such as genomics data, proteomics data, metabolomics data, metagenomics data, transcriptomics data, or other omics data). For example, in certain embodiments, the ML model 704 can include a deep learning model 718, a supervised learning model 720, and an unsupervised learning model 722.

[0095] In certain embodiments, the deep learning model 718 can include any artificial neural network (ANN) that can be used to learn deep representations and abstract concepts from large amounts of data. For example, the deep learning model 718 can include ANNs such as perceptrons, multi-layer perceptrons (MLPs), autoencoders (AEs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), long short-term memory (LSTMs), gated recurrent units (GRUs), restricted Boltzmann machines (RBMs), deep belief networks (DBNs), bidirectional recurrent deep neural networks (BRDNNs), generative adversarial networks (GANs), and deep Q-networks, neural autoregressive distribution estimators (NADEs), adversarial networks (ANs), attention models (AMs), spiking neural networks (SNNs), deep reinforcement learning, etc.

[0096] In certain embodiments, the supervised learning model 720 can include any algorithm that can be used to apply, for example, what has been learned in the past using labeled examples to new data for predicting future events. For example, starting from the analysis of a known training data set, the supervised learning model 720 can generate an inferred function to make predictions about output values. The supervised learning model 720 can also compare its output with the correct and expected output and find errors in order to modify the supervised learning model 720 accordingly. On the other hand, the unsupervised learning model 722 can include, for example, any algorithm that can be applied when the data used to train the unsupervised learning model 722 is neither classified nor labeled. For example, the unsupervised learning model 722 can study and analyze how the system infers a function that describes the hidden structure from unlabeled data.

[0097] In some embodiments, the NLP model 706 can include any algorithm or function that can be adapted to automatically manipulate natural language, such as speech and / or text. For example, in some embodiments, the NLP model 706 can include a content extraction model 724, a classification model 726, a machine translation model 728, a question answering (QA) model 730, and a text generation model 732. In certain embodiments, the content extraction model 724 can include means for extracting text or images from electronic documents (e.g., web pages, text editor documents, etc.) for use in other applications, for example.

[0098] In some embodiments, the classification model 726 can include any algorithm that can learn from data inputs to a supervised learning model (e.g., logistic regression, naive Bayes, stochastic gradient descent (SGD), k-nearest neighbor, decision tree, random forest, support vector machine (SVM), etc.) and make new observations or classifications based thereon. The machine translation model 728 can include any algorithm or function that can be adapted to automatically translate source text in one language into text in another language, for example. The QA model 730 can include any algorithm or function that can be adapted to automatically answer questions posed by humans in natural language, such as those performed by a voice-controlled personal assistant device. The text generation model 732 can include any algorithm or function that can be adapted to automatically generate natural language text.

[0099] In some embodiments, the expert system 708 can include any algorithm or function that can be adapted to simulate the judgment and behavior of a human or organization with expertise and experience in a particular field (e.g., stock trading, medicine, sports statistics, etc.). The computer-based vision model 710 can include any algorithm or function that can be adapted to automatically extract information from images (e.g., photographic images, video images). For example, the computer-based vision model 710 can include an image recognition algorithm 734 and a machine vision algorithm 736. The image recognition algorithm 734 can include any algorithm that can be adapted to automatically identify and / or classify objects, locations, people, etc. that may be included in one or more image frames or other display data, for example. The machine vision algorithm 736 can include any algorithm that can be adapted to allow a computer to "see" or, for example, acquire images using an image sensor camera with dedicated optical elements for processing, analyzing, and / or measuring various data characteristics for decision-making purposes.

[0100] In some embodiments, the speech recognition model 712 can include any algorithm or function that can be suitable for recognizing spoken language and translating it into text, such as through automatic speech recognition (ASR), computer speech recognition, speech-to-text (STT) 738, or text-to-speech (TTS) 740, for example in order to perform calculations to communicate with one or more users via speech. In some embodiments, the planning model 714 can include any algorithm or function that can be suitable for generating a sequence of actions, where each action can include its own set of preconditions to be satisfied before performing that action. Examples of AI planning can include classical planning, reduction to other problems, temporal planning, probabilistic planning, preference-based planning, conditional planning, etc. Finally, the robot model 716 can include any algorithm, function, or system that can enable one or more devices to replicate human behavior, for example through movement, posture, performing tasks, making decisions, emotions, etc.

[0101] As used herein, "or" is inclusive and not exclusive, unless expressly stated otherwise or the context otherwise indicates. Thus, as used herein, "A or B" means "A, B, or both", unless expressly stated otherwise or the context otherwise indicates. Additionally, as used herein, "and" is both conjunctive and disjunctive, unless expressly stated otherwise or the context otherwise indicates. Thus, as used herein, "A and B" means "A and B, jointly or severally", unless expressly stated otherwise or the context otherwise indicates.

[0102] As used herein, "automatic" and its derivatives mean "without human intervention", unless expressly indicated otherwise or the context otherwise indicates.

[0103] The embodiments disclosed herein are merely examples, and the scope of the present disclosure is not limited thereto. The embodiments according to the present disclosure are particularly disclosed in the appended claims directed to methods, storage media, systems, and computer program products, where any feature mentioned in one claim category (e.g., method) can be claimed for protection in another claim category (e.g., system). The dependencies or references in the appended claims are chosen only for form reasons. However, any subject matter resulting from an intentional reference to any of the foregoing claims (in particular multiple dependencies) can also be claimed, and thus any combination of the claims and their features can be disclosed and claimed regardless of the dependencies chosen in the appended claims. The subject matter that can be claimed includes not only combinations of the features listed in the appended claims, but also any other combination of the features in the claims, where each feature mentioned in the claims can be combined with any other feature or combination of features in the claims. Additionally, any embodiment and feature described or depicted herein can be claimed in a separate claim and / or in any combination with any embodiment or feature described or depicted herein or with any feature of the appended claims.

[0104] The scope of the present disclosure encompasses all changes, substitutions, variations, alterations, and modifications to the exemplary embodiments described or illustrated herein that would be understood by a person of ordinary skill in the art. The scope of the present disclosure is not limited to the exemplary embodiments described or illustrated herein. Additionally, although the present disclosure describes and illustrates the corresponding embodiments herein as including specific components, elements, features, functions, operations, or steps, any one of these embodiments may include any combination or arrangement of any components, elements, features, functions, operations, or steps described or illustrated anywhere herein that would be understood by a person of ordinary skill in the art. Further, in the appended claims, a reference to a device or system or a component of a device or system being adapted to, arranged to, capable of, configured to, enabled to, operable to, or operative to perform a particular function encompasses that device, system, component, whether or not the particular function is activated, turned on, or unlocked, so long as the device, system, or component is so adapted, arranged, capable, configured, enabled, operable, or operative. Additionally, although the present disclosure describes or illustrates certain advantages provided by certain embodiments, certain embodiments may not provide, partially provide, or provide all of these advantages.

[0105] Embodiment

[0106] The provided embodiments are:

[0107] 1. A method for identifying a high-reflectivity focus (HRF) in a patient's eye, the method comprising, by one or more computing devices:

[0108] accessing one or more optical coherence tomography (OCT) scans of the retina of the patient's eye;

[0109] inputting the one or more OCT scans into one or more machine learning models that are trained to segment the one or more OCT scans to identify a set of high-reflectivity entities detectable from the one or more OCT scans;

[0110] determining, based on the segmented one or more OCT scans, one or more diameter measurements corresponding to each high-reflectivity entity in the identified set of high-reflectivity entities; and

[0111] identifying a high-reflectivity focus (HRF) in the retina of the patient's eye based on whether at least one of the one or more diameter measurements meets a diameter threshold.

[0112] 2. The method according to embodiment 1, wherein the set of high-reflectivity entities includes a set of high-reflectivity material (HRM), intraretinal high-reflectivity material (IHRM), and HRF.

[0113] 3. The method according to any one of embodiments 1 to 2, wherein identifying the HRF in the retina of the eye of the patient includes identifying a subset of the set of identified highly reflective entities.

[0114] 4. The method according to any one of embodiments 1 to 3, further comprising identifying intraretinal highly reflective material (IHRM) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a second diameter threshold.

[0115] 5. The method according to any one of embodiments 1 to 4, wherein the diameter threshold includes a minimum diameter of approximately 50 micrometers (μm).

[0116] 6. The method according to any one of embodiments 4 to 5, wherein the second diameter threshold includes a diameter range of approximately 50 micrometers (μm) to 100 μm.

[0117] 7. The method according to embodiment 1, wherein determining whether at least one of the one or more diameter measurements meets the diameter threshold further comprises:

[0118] For each highly reflective entity in the set of identified highly reflective entities:

[0119] Associating an ellipse with the identified highly reflective entity;

[0120] Determining the diameter of the longest axis of the ellipse; and

[0121] Estimating at least one of the one or more diameter measurements based on the diameter of the longest axis of the ellipse.

[0122] 8. The method according to embodiment 1, wherein identifying the HRF in the retina of the eye of the patient further comprises classifying the eye of the patient as having at least one of the following: diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), or retinal vein occlusion (RVO).

[0123] 9. The method according to embodiment 1, wherein identifying the HRF in the retina of the eye of the patient further comprises:

[0124] Accessing ETDRS grid mapping information identifying one or more partitions of an Early Treatment Diabetic Retinopathy Study (ETDRS) grid; and

[0125] Determine one or more volume measurements of the identified HRF at least in part based on the ETDRS grid mapping information.

[0126] 10. The method according to embodiment 9, wherein the one or more volume measurements include one or more of the following: the HRF volume in the retina of the patient's eye, the HRF area in the retina of the patient's eye, or the HRF thickness in the retina of the patient's eye.

[0127] 11. The method according to any one of embodiments 9 to 10, further comprising determining the HRF volume in the retina of the patient's eye relative to at least one of the identified one or more partitions at least in part based on the ETDRS grid mapping information.

[0128] 12. The method according to embodiment 11, wherein determining the HRF volume in the retina of the patient's eye includes determining a decrease in the HRF volume in the retina of the patient's eye.

[0129] 13. The method according to embodiment 11, wherein the at least one of the identified one or more partitions includes the outer retina partition of the ETDRS grid.

[0130] 14. The method according to any one of embodiments 9 to 13, further comprising determining the number of HRFs in the retina of the patient's eye relative to at least one of the identified one or more partitions at least in part based on the ETDRS grid mapping information.

[0131] 15. The method according to any one of embodiments 9 to 14, further comprising:

[0132] Access a frontal image of the retina of the patient's eye, wherein the frontal image is associated with the one or more OCT scans; and

[0133] Map the identified HRF to the frontal image at least in part based on the ETDRS grid mapping information.

[0134] 16. The method according to embodiment 1, wherein the one or more machine learning models include at least one semantic segmentation model.

[0135] 17. The method according to embodiment 16, wherein the at least one semantic segmentation model includes a U-Net architecture.

[0136] 18. The method according to embodiment 1, further comprising training the one or more machine learning models by:

[0137] Accessing an OCT scan data set of the retinas of the eyes of one or more patients, wherein the OCT scan data set includes sparse annotations of HRFs in the retinas of the eyes of the one or more patients;

[0138] Dividing the OCT scan data set into a model training data set and a model validation data set;

[0139] Training the one or more machine learning models based on the model training data set to segment OCT scans to identify a set of highly reflective entities detectable from the OCT scans, wherein the identified set of highly reflective entities includes highly reflective material (HRM); and

[0140] Evaluating the one or more machine learning models based on the model validation data set.

[0141] 19. The method according to embodiment 18, further comprising identifying HRFs in the retinas of the eyes of the one or more patients based on whether one or more diameter measurements of the HRM meet a predetermined diameter threshold.

[0142] 20. The method according to embodiment 18, wherein the sparse annotation of the HRF includes a boundary geometry surrounding a plurality of HRF instances, and the method further comprises:

[0143] Before training the one or more machine learning models, adjusting the boundary geometry by reducing the size of the boundary geometry, the size of the boundary geometry being reduced to annotate a single HRF instance.

[0144] 21. The method according to embodiment 1, wherein the one or more OCT scans include one or more first OCT scans of the retinas of the eyes of the patient captured on an initial date, and wherein the identified HRF includes a first HRF volume, and the method further comprises:

[0145] Accessing one or more second OCT scans of the retinas of the eyes of the patient;

[0146] Inputting the one or more second OCT scans into the one or more machine learning models to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans;

[0147] Determine one or more second diameter measurements corresponding to each of the highly reflective entities in the identified second set of highly reflective entities, based on the one or more segmented second OCT scans;

[0148] Identify a second HRF volume in the retina of the eye of the patient based on whether at least one of the one or more second diameter measurements meets the diameter threshold; and

[0149] Determine whether the eye of the patient responds to the treatment based on the second HRF volume.

[0150] 22. The method according to embodiment 21, further comprising determining the degree to which the eye of the patient responds to the treatment based on the second HRF volume.

[0151] 23. The method according to any one of embodiments 21 to 22, wherein when the second HRF volume is less than the first HRF volume, the eye of the patient responds to the treatment.

[0152] 24. The method according to any one of embodiments 21 to 23, wherein the treatment comprises an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof.

[0153] 25. The method according to any one of embodiments 21 to 24, wherein the anti-VEGF-A antibody comprises faricimab.

[0154] 26. The method according to any one of embodiments 21 to 25, wherein the anti-Ang-2 antibody comprises faricimab.

[0155] 27. The method according to any one of embodiments 21 to 26, wherein the anti-VEGF antibody is selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium.

[0156] 28. The method according to embodiment 1, further comprising identifying an effective treatment regimen of an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof to treat the eye of the patient based on the volume or number of the identified HRF.

[0157] 29. The method according to embodiment 28, wherein identifying the effective treatment regimen comprises identifying a dose for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRF.

[0158] 30. The method according to any one of embodiments 28 to 29, wherein identifying the effective treatment regimen includes identifying a plan for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

[0159] 31. The method according to any one of embodiments 28 to 30, wherein identifying the effective treatment regimen includes identifying a duration for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

[0160] 32. The method according to embodiment 21, wherein the one or more second OCT scans are captured on one or more dates selected from the group consisting of: about 0.25 months, 0.5 months, 0.75 months, 1 month, 3 months, 6 months, 9 months, 12 months, 15 months, 18 months, 21 months, 24 months, 27 months, 30 months, 33 months, or 36 months from the initial date.

[0161] 33. A system for identifying high-reflectivity foci (HRFs) in a patient's eye, the system comprising one or more computing devices, the one or more computing devices comprising:

[0162] One or more non-transitory computer-readable storage media comprising instructions; and

[0163] One or more processors coupled to the one or more storage media, the one or more processors being configured to execute the instructions to:

[0164] Access one or more optical coherence tomography (OCT) scans of the retina of the patient's eye;

[0165] Input the one or more OCT scans into one or more machine learning models, the one or more machine learning models being trained to segment the one or more OCT scans to identify a set of high-reflectivity entities detectable from the one or more OCT scans;

[0166] Based on the segmented one or more OCT scans, determine one or more diameter measurements corresponding to each high-reflectivity entity in the identified set of high-reflectivity entities; and

[0167] Identify high-reflectivity foci (HRFs) in the retina of the patient's eye based on whether at least one of the one or more diameter measurements meets a diameter threshold.

[0168] 34. The system according to embodiment 33, wherein the set of highly reflective entities includes a set of highly reflective material (HRM), intraretinal highly reflective material (IHRM), and HRF.

[0169] 35. The system according to any one of embodiments 33 to 34, wherein the instructions for identifying HRF in the retina of the eye of the patient further include instructions for identifying a subset of the identified set of highly reflective entities.

[0170] 36. The system according to any one of embodiments 33 to 35, wherein the instructions further include instructions for identifying intraretinal highly reflective material (IHRM) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a second diameter threshold.

[0171] 37. The system according to any one of embodiments 33 to 36, wherein the diameter threshold includes a minimum diameter of approximately 50 micrometers (μm).

[0172] 38. The system according to any one of embodiments 36 to 37, wherein the second diameter threshold includes a diameter range of approximately 50 micrometers (μm) to 100 μm.

[0173] 39. The system according to embodiment 33, wherein the instructions for determining whether at least one of the one or more diameter measurements meets the diameter threshold further include instructions for performing the following:

[0174] For each highly reflective entity in the identified set of highly reflective entities:

[0175] Associate an ellipse with the identified highly reflective entity;

[0176] Determine the diameter of the longest axis of the ellipse; and

[0177] Estimate at least one of the one or more diameter measurements based on the diameter of the longest axis of the ellipse.

[0178] 40. The system according to embodiment 33, wherein the instructions for identifying HRF in the retina of the eye of the patient further include instructions for classifying the eye of the patient as having at least one of the following: diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), or retinal vein occlusion (RVO).

[0179] 41. The system according to embodiment 33, wherein the instructions for identifying the HRF in the retina of the eye of the patient further include instructions for performing the following:

[0180] Accessing ETDRS grid mapping information that identifies one or more partitions of an Early Treatment Diabetic Retinopathy Study (ETDRS) grid; and

[0181] Determining one or more volume measurements of the identified HRF based at least in part on the ETDRS grid mapping information.

[0182] 42. The system according to embodiment 41, wherein the one or more volume measurements include one or more of the following: the HRF volume in the retina of the eye of the patient, the HRF area in the retina of the eye of the patient, or the HRF thickness in the retina of the eye of the patient.

[0183] 43. The system according to any one of embodiments 41 to 42, wherein the instructions further include instructions for determining, based at least in part on the ETDRS grid mapping information, the HRF volume in the retina of the eye of the patient relative to at least one of the identified one or more partitions.

[0184] 44. The system according to embodiment 43, wherein the instructions for determining the HRF volume in the retina of the eye of the patient further include instructions for determining a decrease in the HRF volume in the retina of the eye of the patient.

[0185] 45. The system according to embodiment 43, wherein at least one of the identified one or more partitions includes an outer retina partition of the ETDRS grid.

[0186] 46. The system according to any one of embodiments 41 to 45, wherein the instructions further include instructions for determining, based at least in part on the ETDRS grid mapping information, the number of HRFs in the retina of the eye of the patient relative to at least one of the identified one or more partitions.

[0187] 47. The system according to any one of embodiments 41 to 46, wherein the instructions further include instructions for performing the following:

[0188] Accessing a frontal image of the retina of the eye of the patient, wherein the frontal image is associated with the one or more OCT scans; and

[0189] Map the identified HRFs to the frontal image based at least in part on the ETDRS grid mapping information.

[0190] 48. The system according to embodiment 33, wherein the one or more machine learning models include at least one semantic segmentation model.

[0191] 49. The system according to embodiment 48, wherein the at least one semantic segmentation model includes a U-Net architecture.

[0192] 50. The system according to embodiment 33, wherein the instructions further include instructions for training the one or more machine learning models by:

[0193] Accessing an OCT scan data set of the retinas of the eyes of one or more patients, wherein the OCT scan data set includes sparse annotations of HRFs in the retinas of the eyes of the one or more patients;

[0194] Partitioning the OCT scan data set into a model training data set and a model validation data set;

[0195] Training the one or more machine learning models based on the model training data set to segment OCT scans to identify a set of highly reflective entities detectable from the OCT scans, wherein the identified set of highly reflective entities includes highly reflective material (HRM); and

[0196] Evaluating the one or more machine learning models based on the model validation data set.

[0197] 51. The system according to embodiment 50, wherein the instructions further include instructions for identifying HRFs in the retinas of the eyes of the one or more patients based on whether one or more diameter measurements of the HRM meet a predetermined diameter threshold.

[0198] 52. The system according to embodiment 50, wherein the sparse annotation of the HRF includes a boundary geometry enclosing a plurality of HRF instances, and wherein the instructions further include instructions for:

[0199] Before training the one or more machine learning models, adjusting the boundary geometry by reducing the size of the boundary geometry, the size of the boundary geometry being reduced to annotate a single HRF instance.

[0200] 53. The system according to embodiment 33, wherein the one or more OCT scans include one or more first OCT scans of the retina of the eye of the patient captured on an initial date, wherein the identified HRF includes a first HRF volume, and wherein the instructions further include instructions for:

[0201] accessing one or more second OCT scans of the retina of the eye of the patient;

[0202] inputting the one or more second OCT scans into the one or more machine learning models to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans;

[0203] determining, based on the segmented one or more second OCT scans, one or more second diameter measurements corresponding to each highly reflective entity in the identified second set of highly reflective entities;

[0204] identifying a second HRF volume in the retina of the eye of the patient based on whether at least one of the one or more second diameter measurements meets the diameter threshold; and

[0205] determining whether the eye of the patient responds to the treatment based on the second HRF volume.

[0206] 54. The system according to embodiment 53, wherein the instructions further include instructions for determining the degree to which the eye of the patient responds to the treatment based on the second HRF volume.

[0207] 55. The system according to any one of embodiments 53 to 54, wherein the eye of the patient responds to the treatment when the second HRF volume is less than the first HRF volume.

[0208] 56. The system according to any one of embodiments 53 to 55, wherein the treatment includes an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof.

[0209] 57. The system according to any one of embodiments 53 to 56, wherein the anti-VEGF-A antibody includes faricimab.

[0210] 58. The system according to any one of embodiments 53 to 57, wherein the anti-Ang-2 antibody includes faricimab.

[0211] 59. The system according to any one of embodiments 53 to 58, wherein the anti-VEGF antibody is selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium.

[0212] 60. The system according to embodiment 33, wherein the instructions further comprise instructions for identifying an effective treatment regimen for an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof to treat the eye of the patient based on the volume or number of the identified HRF.

[0213] 61. The system according to embodiment 60, wherein the instructions for identifying the effective treatment regimen further comprise instructions for identifying a dose for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRF.

[0214] 62. The system according to any one of embodiments 60 to 61, wherein the instructions for identifying the effective treatment regimen further comprise instructions for identifying a schedule for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRF.

[0215] 63. The system according to any one of embodiments 60 to 62, wherein the instructions for identifying the effective treatment regimen further comprise instructions for identifying a duration for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRF.

[0216] 64. The system according to embodiment 53, wherein the one or more second OCT scans are captured on one or more dates selected from the group consisting of approximately 0.25 months, 0.5 months, 0.75 months, 1 month, 3 months, 6 months, 9 months, 12 months, 15 months, 18 months, 21 months, 24 months, 27 months, 30 months, 33 months, or 36 months from the initial date.

[0217] 65. A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors of one or more computing devices, cause the one or more processors to:

[0218] access one or more optical coherence tomography (OCT) scans of the retina of an eye of a patient;

[0219] Input the one or more OCT scans into one or more machine learning models, the one or more machine learning models being trained to segment the one or more OCT scans to identify a set of highly reflective entities detectable from the one or more OCT scans;

[0220] Based on the segmented one or more OCT scans, determine one or more diameter measurements corresponding to each highly reflective entity in the identified set of highly reflective entities; and

[0221] Identify a highly reflective focus (HRF) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a diameter threshold.

[0222] 66. The non - transitory computer - readable medium according to embodiment 65, wherein the set of highly reflective entities includes a set of highly reflective material (HRM), intra - retinal highly reflective material (IHRM), and HRF.

[0223] 67. The non - transitory computer - readable medium according to any one of embodiments 65 to 66, wherein the instructions for identifying the HRF in the retina of the eye of the patient further include instructions for identifying a subset of the identified set of highly reflective entities.

[0224] 68. The non - transitory computer - readable medium according to any one of embodiments 65 to 67, wherein the instructions further include instructions for identifying intra - retinal highly reflective material (IHRM) in the retina of the eye of the patient based on whether the at least one of the one or more diameter measurements meets a second diameter threshold.

[0225] 69. The non - transitory computer - readable medium according to any one of embodiments 65 to 68, wherein the diameter threshold includes a minimum diameter of approximately 50 micrometers (μm).

[0226] 70. The non - transitory computer - readable medium according to any one of embodiments 68 to 69, wherein the second diameter threshold includes a diameter range of approximately 50 micrometers (μm) to 100 μm.

[0227] 71. The non - transitory computer - readable medium according to embodiment 65, wherein the instructions for determining whether the at least one of the one or more diameter measurements meets the diameter threshold further include instructions for performing the following:

[0228] For each highly reflective entity in the identified set of highly reflective entities:

[0229] Associate an ellipse with the identified highly reflective entity;

[0230] Determine the diameter of the longest axis of the ellipse; and

[0231] Estimate at least one of the one or more diameter measurements based on the diameter of the longest axis of the ellipse.

[0232] 72. The non-transitory computer-readable medium according to embodiment 65, wherein the instructions for identifying the HRF in the retina of the eye of the patient further include instructions for classifying the eye of the patient as having at least one of the following: diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), or retinal vein occlusion (RVO).

[0233] 73. The non-transitory computer-readable medium according to embodiment 65, wherein the instructions for identifying the HRF in the retina of the eye of the patient further include instructions for performing the following:

[0234] Access ETDRS grid mapping information that identifies one or more partitions of an Early Treatment Diabetic Retinopathy Study (ETDRS) grid; and

[0235] Determine one or more volume measurements of the identified HRF at least in part based on the ETDRS grid mapping information.

[0236] 74. The non-transitory computer-readable medium according to embodiment 73, wherein the one or more volume measurements include one or more of the following: the HRF volume in the retina of the eye of the patient, the HRF area in the retina of the eye of the patient, or the HRF thickness in the retina of the eye of the patient.

[0237] 75. The non-transitory computer-readable medium according to any one of embodiments 73 to 74, wherein the instructions further include instructions for determining the HRF volume in the retina of the eye of the patient relative to at least one of the identified one or more partitions at least in part based on the ETDRS grid mapping information.

[0238] 76. The non-transitory computer-readable medium according to embodiment 75, wherein the instructions for determining the HRF volume in the retina of the eye of the patient further include instructions for determining a decrease in the HRF volume in the retina of the eye of the patient.

[0239] 77. The non-transitory computer-readable medium according to embodiment 75, wherein at least one of the one or more identified partitions includes an outer retina partition of the ETDRS grid.

[0240] 78. The non-transitory computer-readable medium according to any one of embodiments 73 to 77, wherein the instructions further include instructions for determining, at least in part based on the ETDRS grid mapping information, the number of HRFs in the retina of the patient's eye relative to at least one of the one or more identified partitions.

[0241] 79. The non-transitory computer-readable medium according to any one of embodiments 73 to 78, wherein the instructions further include instructions for performing the following:

[0242] Accessing a frontal image of the retina of the patient's eye, wherein the frontal image is associated with the one or more OCT scans; and

[0243] Mapping the identified HRFs to the frontal image at least in part based on the ETDRS grid mapping information.

[0244] 80. The non-transitory computer-readable medium according to embodiment 65, wherein the one or more machine learning models include at least one semantic segmentation model.

[0245] 81. The non-transitory computer-readable medium according to embodiment 80, wherein the at least one semantic segmentation model includes a U-Net architecture.

[0246] 82. The non-transitory computer-readable medium according to embodiment 65, wherein the instructions further include instructions for training the one or more machine learning models by:

[0247] Accessing an OCT scan data set of the retinas of the eyes of one or more patients, wherein the OCT scan data set includes sparse annotations of HRFs in the retinas of the eyes of the one or more patients;

[0248] Dividing the OCT scan data set into a model training data set and a model validation data set;

[0249] Training the one or more machine learning models based on the model training data set to segment OCT scans to identify a set of highly reflective entities detectable from the OCT scans, wherein the identified set of highly reflective entities includes highly reflective material (HRM); and

[0250] Evaluate the one or more machine learning models based on the model validation dataset.

[0251] 83. The non-transitory computer-readable medium according to embodiment 82, wherein the instructions further include instructions for identifying HRF in the retina of the eyes of the one or more patients based on whether one or more diameter measurements of the HRM satisfy a predetermined diameter threshold.

[0252] 84. The non-transitory computer-readable medium according to embodiment 82, wherein the sparse annotation of the HRF includes a boundary geometry surrounding a plurality of HRF instances, and wherein the instructions further include instructions for performing the following:

[0253] Before training the one or more machine learning models, adjust the boundary geometry by reducing the size of the boundary geometry, and the size of the boundary geometry is reduced to annotate a single HRF instance.

[0254] 85. The non-transitory computer-readable medium according to embodiment 65, wherein the one or more OCT scans include one or more first OCT scans of the retina of the patient's eye captured on an initial date, wherein the identified HRF includes a first HRF volume, and wherein the instructions further include instructions for performing the following:

[0255] Access one or more second OCT scans of the retina of the patient's eye;

[0256] Input the one or more second OCT scans into the one or more machine learning models to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans;

[0257] Based on the segmented one or more second OCT scans, determine one or more second diameter measurements corresponding to each highly reflective entity in the identified second set of highly reflective entities;

[0258] Based on whether at least one of the one or more second diameter measurements satisfies the diameter threshold, identify a second HRF volume in the retina of the patient's eye; and

[0259] Based on the second HRF volume, determine whether the patient's eye responds to treatment.

[0260] 86. The non-transitory computer-readable medium according to embodiment 85, wherein the instructions further include instructions for determining the degree to which the patient's eye responds to the treatment based on the second HRF volume.

[0261] 87. The non-transitory computer-readable medium according to any one of embodiments 85 to 86, wherein the patient's eye responds to the treatment when the second HRF volume is less than the first HRF volume.

[0262] 88. The non-transitory computer-readable medium according to any one of embodiments 85 to 87, wherein the treatment includes an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof.

[0263] 89. The non-transitory computer-readable medium according to embodiments 85 to 88, wherein the anti-VEGF-A antibody includes faricimab.

[0264] 90. The non-transitory computer-readable medium according to embodiments 85 to 89, wherein the anti-Ang-2 antibody includes faricimab.

[0265] 91. The non-transitory computer-readable medium according to embodiments 85 to 90, wherein the anti-VEGF antibody is selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium.

[0266] 92. The non-transitory computer-readable medium according to embodiment 65, wherein the instructions further include instructions for identifying an effective treatment regimen for an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof to treat the patient's eye based on the volume or number of the identified HRF.

[0267] 93. The non-transitory computer-readable medium according to embodiment 92, wherein the instructions for identifying the effective treatment regimen further include instructions for identifying a dose for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or number of the identified HRF.

[0268] 94. The non-transitory computer-readable medium according to any one of embodiments 92 to 93, wherein the instructions for identifying the effective treatment regimen further include instructions for identifying a schedule for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or number of the identified HRF.

[0269] 95. The non-transitory computer-readable medium according to any one of embodiments 92 to 94, wherein the instructions for identifying the effective treatment regimen further include instructions for identifying a duration for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the quantity of the identified HRF.

[0270] 96. The non-transitory computer-readable medium according to embodiment 85, wherein the one or more second OCT scans are captured on one or more dates selected from the group consisting of: about 0.25 months, 0.5 months, 0.75 months, 1 month, 3 months, 6 months, 9 months, 12 months, 15 months, 18 months, 21 months, 24 months, 27 months, 30 months, 33 months, or 36 months from the initial date.

[0271] 97. The method according to any one of embodiments 1 to 32, further comprising: determining whether at least one of the one or more diameter measurements meets a third diameter threshold.

[0272] 98. The method according to embodiment 97, wherein the third diameter threshold comprises a minimum diameter of about 100 micrometers (μm).

[0273] 99. The system according to any one of embodiments 33 to 64, further comprising: determining whether at least one of the one or more diameter measurements meets a third diameter threshold.

[0274] 100. The system according to embodiment 99, wherein the third diameter threshold comprises a minimum diameter of about 100 micrometers (μm).

[0275] 101. The non-transitory computer-readable medium according to any one of embodiments 65 to 96, further comprising: determining whether at least one of the one or more diameter measurements meets a third diameter threshold.

[0276] 102. The non-transitory computer-readable medium according to embodiment 101, wherein the third diameter threshold comprises a minimum diameter of about 100 micrometers (μm).

Claims

1. A method for identifying high - reflective foci (HRF) in a patient's eye, the method comprising, by one or more computing devices: accessing one or more optical coherence tomography (OCT) scans of the retina of the patient's eye; inputting the one or more OCT scans into one or more machine - learning models, the one or more machine - learning models being trained to segment the one or more OCT scans to identify a set of high - reflective entities detectable from the one or more OCT scans; determining, based on the segmented one or more OCT scans, one or more diameter measurements corresponding to each high - reflective entity in the identified set of high - reflective entities; and identifying a high - reflective focus (HRF) in the retina of the patient's eye based on whether at least one of the one or more diameter measurements meets a diameter threshold.

2. The method according to claim 1, wherein the set of high - reflective entities includes a set of high - reflective material (HRM), intra - retinal high - reflective material (IHRM), and HRF.

3. The method according to any one of claims 1 to 2, wherein identifying the HRF in the retina of the patient's eye includes identifying a subset of the identified set of high - reflective entities.

4. The method according to any one of claims 1 to 3, further comprising identifying intra - retinal high - reflective material (IHRM) in the retina of the patient's eye based on whether the at least one of the one or more diameter measurements meets a second diameter threshold.

5. The method according to any one of claims 1 to 4, wherein the diameter threshold includes a minimum diameter of approximately 50 micrometers (μm).

6. The method according to any one of claims 4 to 5, wherein the second diameter threshold includes a diameter range of approximately 50 micrometers (μm) to 100 μm.

7. The method according to claim 1, wherein determining whether at least one of the one or more diameter measurements meets the diameter threshold further comprises: for each high - reflective entity in the identified set of high - reflective entities: associating an ellipse with the identified high - reflective entity; determining the diameter of the longest axis of the ellipse; and estimating at least one of the one or more diameter measurements based on the diameter of the longest axis of the ellipse.

8. The method according to claim 1, wherein identifying the HRF in the retina of the patient's eye further comprises classifying the patient's eye as having at least one of the following: diabetic retinopathy (DR), diabetic macular edema (DME), age - related macular degeneration (AMD), neovascular age - related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), or retinal vein occlusion (RVO).

9. The method according to claim 1, wherein identifying the HRF in the retina of the patient's eye further comprises: Access ETDRS grid mapping information that identifies one or more partitions of an Early Treatment Diabetic Retinopathy Study (ETDRS) grid; and Determine one or more volume measurements of the identified HRF based at least in part on the ETDRS grid mapping information.

10. The method of claim 9, wherein the one or more volume measurements include one or more of the following: the HRF volume in the retina of the patient's eye, the HRF area in the retina of the patient's eye, or the HRF thickness in the retina of the patient's eye.

11. The method of any one of claims 9 to 10, further comprising determining the HRF volume in the retina of the patient's eye relative to at least one of the identified one or more partitions based at least in part on the ETDRS grid mapping information.

12. The method of claim 11, wherein determining the HRF volume in the retina of the patient's eye includes determining a decrease in the HRF volume in the retina of the patient's eye.

13. The method of claim 11, wherein the at least one of the identified one or more partitions includes an outer retina partition of the ETDRS grid.

14. The method of any one of claims 9 to 13, further comprising determining the number of HRFs in the retina of the patient's eye relative to at least one of the identified one or more partitions based at least in part on the ETDRS grid mapping information.

15. The method of any one of claims 9 to 14, further comprising: Access a front view image of the retina of the patient's eye, wherein the front view image is associated with the one or more OCT scans; and Map the identified HRF to the front view image based at least in part on the ETDRS grid mapping information.

16. The method of claim 1, wherein the one or more machine learning models include at least one semantic segmentation model.

17. The method of claim 16, wherein the at least one semantic segmentation model includes a U-Net architecture.

18. The method of claim 1, further comprising training the one or more machine learning models by: Accessing an OCT scan data set of the retinas of the eyes of one or more patients, wherein the OCT scan data set includes sparse annotations of HRFs in the retinas of the eyes of the one or more patients; Partitioning the OCT scan data set into a model training data set and a model validation data set; Training the one or more machine learning models based on the model training data set to segment OCT scans to identify a set of highly reflective entities detectable from the OCT scans, wherein the identified set of highly reflective entities includes highly reflective material (HRM); and Evaluate the one or more machine learning models based on the model validation dataset.

19. The method according to claim 18, further comprising identifying the HRF in the retina of the eye of the one or more patients based on whether one or more diameter measurements of the HRM meet a predetermined diameter threshold.

20. The method according to claim 18, wherein the sparse annotation of the HRF includes a boundary geometry surrounding a plurality of HRF instances, and the method further comprises: Before training the one or more machine learning models, adjusting the boundary geometry by reducing the size of the boundary geometry, the size of the boundary geometry being reduced to annotate a single HRF instance.

21. The method according to claim 1, wherein the one or more OCT scans include one or more first OCT scans of the retina of the eye of the patient captured on an initial date, and wherein the identified HRF includes a first HRF volume, and the method further comprises: Accessing one or more second OCT scans of the retina of the eye of the patient; Inputting the one or more second OCT scans into the one or more machine learning models to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans; Based on the segmented one or more second OCT scans, determining one or more second diameter measurements corresponding to each highly reflective entity in the identified second set of highly reflective entities; Identifying a second HRF volume in the retina of the eye of the patient based on whether at least one of the one or more second diameter measurements meets the diameter threshold; and Determining whether the eye of the patient responds to the treatment based on the second HRF volume.

22. The method according to claim 21, further comprising determining the degree to which the eye of the patient responds to the treatment based on the second HRF volume.

23. The method according to any one of claims 21 to 22, wherein the eye of the patient responds to the treatment when the second HRF volume is less than the first HRF volume.

24. The method according to any one of claims 21 to 23, wherein the treatment comprises an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof.

25. The method according to claims 21 to 24, wherein the anti-VEGF-A antibody comprises faricimab-svoa.

26. The method according to claims 21 to 25, wherein the anti-Ang-2 antibody comprises faricimab.

27. The method according to any one of claims 21 to 26, wherein the anti-VEGF antibody is selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium.

28. The method according to claim 1, further comprising identifying an effective treatment regimen for an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof to treat the eye of the patient based on the volume or number of the identified HRFs.

29. The method according to claim 28, wherein identifying the effective treatment regimen comprises identifying a dose for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

30. The method according to any one of claims 28 to 29, wherein identifying the effective treatment regimen comprises identifying a schedule for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

31. The method according to any one of claims 28 to 30, wherein identifying the effective treatment regimen comprises identifying a duration for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

32. The method according to claim 21, wherein the one or more second OCT scans are captured on one or more dates selected from the group consisting of approximately 0.25 months, 0.5 months, 0.75 months, 1 month, 3 months, 6 months, 9 months, 12 months, 15 months, 18 months, 21 months, 24 months, 27 months, 30 months, 33 months, or 36 months from the initial date.

33. A system for identifying high-reflectivity foci (HRFs) in the eye of a patient, the system comprising one or more computing devices, the one or more computing devices comprising: one or more non-transitory computer-readable storage media comprising instructions; and one or more processors coupled to the one or more storage media, the one or more processors being configured to execute the instructions to: access one or more optical coherence tomography (OCT) scans of the retina of the eye of the patient; input the one or more OCT scans into one or more machine learning models, the one or more machine learning models being trained to segment the one or more OCT scans to identify a set of high-reflectivity entities detectable from the one or more OCT scans; determine, based on the segmented one or more OCT scans, one or more diameter measurements corresponding to each high-reflectivity entity in the identified set of high-reflectivity entities; and identify high-reflectivity foci (HRFs) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a diameter threshold.

34. The system according to claim 33, wherein the set of highly reflective entities includes a set of highly reflective material (HRM), intraretinal highly reflective material (IHRM), and HRF.

35. The system according to any one of claims 33 to 34, wherein the instructions for identifying the HRF in the retina of the eye of the patient further include instructions for identifying a subset of the identified set of highly reflective entities.

36. The system according to any one of claims 33 to 35, wherein the instructions further include instructions for identifying intraretinal highly reflective material (IHRM) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a second diameter threshold.

37. The system according to any one of claims 33 to 36, wherein the diameter threshold includes a minimum diameter of approximately 50 micrometers (μm).

38. The system according to any one of claims 36 to 37, wherein the second diameter threshold includes a diameter range of approximately 50 micrometers (μm) to 100 μm.

39. The system according to claim 33, wherein the instructions for determining whether at least one of the one or more diameter measurements meets the diameter threshold further include instructions for performing the following: For each highly reflective entity in the identified set of highly reflective entities: Associate an ellipse with the identified highly reflective entity; Determine the diameter of the longest axis of the ellipse; and Estimate at least one of the one or more diameter measurements based on the diameter of the longest axis of the ellipse.

40. The system according to claim 33, wherein the instructions for identifying the HRF in the retina of the eye of the patient further include instructions for classifying the eye of the patient as having at least one of the following: diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), or retinal vein occlusion (RVO).

41. The system according to claim 33, wherein the instructions for identifying the HRF in the retina of the eye of the patient further include instructions for performing the following: Access ETDRS grid mapping information identifying one or more partitions of an Early Treatment Diabetic Retinopathy Study (ETDRS) grid; and Determine one or more volume measurements of the identified HRF at least in part based on the ETDRS grid mapping information.

42. The system according to claim 41, wherein the one or more volume measurements include one or more of the following: the HRF volume in the retina of the eye of the patient, the HRF area in the retina of the eye of the patient, or the HRF thickness in the retina of the eye of the patient.

43. The system according to any one of claims 41 to 42, wherein the instructions further include instructions for determining the HRF volume in the retina of the patient's eye relative to at least one of the identified one or more partitions based at least in part on the ETDRS grid mapping information.

44. The system according to claim 43, wherein the instructions for determining the HRF volume in the retina of the patient's eye further include instructions for determining a reduction in the HRF volume in the retina of the patient's eye.

45. The system according to claim 43, wherein at least one of the identified one or more partitions includes an outer retina partition of the ETDRS grid.

46. The system according to any one of claims 41 to 45, wherein the instructions further include instructions for determining the number of HRFs in the retina of the patient's eye relative to at least one of the identified one or more partitions based at least in part on the ETDRS grid mapping information.

47. The system according to any one of claims 41 to 46, wherein the instructions further include instructions for performing the following: accessing a frontal image of the retina of the patient's eye, wherein the frontal image is associated with the one or more OCT scans; and mapping the identified HRFs to the frontal image based at least in part on the ETDRS grid mapping information.

48. The system according to claim 33, wherein the one or more machine learning models include at least one semantic segmentation model.

49. The system according to claim 48, wherein the at least one semantic segmentation model includes a U-Net architecture.

50. The system according to claim 33, wherein the instructions further include instructions for training the one or more machine learning models by: accessing an OCT scan data set of the retinas of the eyes of one or more patients, wherein the OCT scan data set includes sparse annotations of HRFs in the retinas of the eyes of the one or more patients; dividing the OCT scan data set into a model training data set and a model validation data set; training the one or more machine learning models based on the model training data set to segment OCT scans to identify a set of highly reflective entities detectable from the OCT scans, wherein the identified set of highly reflective entities includes highly reflective material (HRM); and evaluating the one or more machine learning models based on the model validation data set.

51. The system according to claim 50, wherein the instructions further include instructions for identifying HRFs in the retinas of the eyes of the one or more patients based on whether one or more diameter measurements of the HRM meet a predetermined diameter threshold.

52. The system according to claim 50, wherein the sparse annotation of the HRF includes boundary geometry surrounding a plurality of HRF instances, and wherein the instructions further include instructions for performing the following: Before training the one or more machine learning models, adjusting the boundary geometry by reducing the size of the boundary geometry, the size of the boundary geometry being reduced to annotate a single HRF instance.

53. The system according to claim 33, wherein the one or more OCT scans include one or more first OCT scans of the retina of the patient's eye captured at an initial date, wherein the identified HRF includes a first HRF volume, and wherein the instructions further include instructions for performing the following: Accessing one or more second OCT scans of the retina of the patient's eye; Inputting the one or more second OCT scans into the one or more machine learning models to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans; Based on the segmented one or more second OCT scans, determining one or more second diameter measurements corresponding to each highly reflective entity in the identified second set of highly reflective entities; Identifying a second HRF volume in the retina of the patient's eye based on whether at least one of the one or more second diameter measurements meets the diameter threshold; and Determining whether the patient's eye responds to treatment based on the second HRF volume.

54. The system according to claim 53, wherein the instructions further include instructions for determining the degree to which the patient's eye responds to the treatment based on the second HRF volume.

55. The system according to any one of claims 53 to 54, wherein the patient's eye responds to the treatment when the second HRF volume is less than the first HRF volume.

56. The system according to any one of claims 53 to 55, wherein the treatment includes an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof.

57. The system according to claims 53 to 56, wherein the anti-VEGF-A antibody includes faricimab.

58. The system according to claims 53 to 57, wherein the anti-Ang-2 antibody includes faricimab.

59. The system according to claims 53 to 58, wherein the anti-VEGF antibody is selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium.

60. The system according to claim 33, wherein the instructions further include instructions for identifying an effective treatment regimen for an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof to treat the eye of the patient based on the volume or number of the identified HRFs.

61. The system according to claim 60, wherein the instructions for identifying the effective treatment regimen further include instructions for identifying a dose for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

62. The system according to any one of claims 60 to 61, wherein the instructions for identifying the effective treatment regimen further include instructions for identifying a schedule for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

63. The system according to any one of claims 60 to 62, wherein the instructions for identifying the effective treatment regimen further include instructions for identifying a duration for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRFs.

64. The system according to claim 53, wherein the one or more second OCT scans are captured on one or more dates selected from the group consisting of: about 0.25 months, 0.5 months, 0.75 months, 1 month, 3 months, 6 months, 9 months, 12 months, 15 months, 18 months, 21 months, 24 months, 27 months, 30 months, 33 months, or 36 months from the initial date.

65. A non-transitory computer-readable medium comprising instructions that, when executed by one or more processors of one or more computing devices, cause the one or more processors to: access one or more optical coherence tomography (OCT) scans of the retina of an eye of a patient; input the one or more OCT scans into one or more machine learning models that are trained to segment the one or more OCT scans to identify a set of hyper-reflective entities detectable from the one or more OCT scans; determine, based on the segmented one or more OCT scans, one or more diameter measurements corresponding to each hyper-reflective entity in the identified set of hyper-reflective entities; and identify a hyper-reflective focus (HRF) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a diameter threshold.

66. The non-transitory computer-readable medium according to claim 65, wherein the set of hyper-reflective entities includes a set of hyper-reflective material (HRM), intraretinal hyper-reflective material (IHRM), and HRFs.

67. The non-transitory computer-readable medium according to any one of claims 65 to 66, wherein the instructions for identifying the HRFs in the retina of the eye of the patient further include instructions for identifying a subset of the identified set of highly reflective entities.

68. The non-transitory computer-readable medium according to any one of claims 65 to 67, wherein the instructions further include instructions for identifying intraretinal highly reflective material (IHRM) in the retina of the eye of the patient based on whether at least one of the one or more diameter measurements meets a second diameter threshold.

69. The non-transitory computer-readable medium according to any one of claims 65 to 68, wherein the diameter threshold includes a minimum diameter of approximately 50 micrometers (μm).

70. The non-transitory computer-readable medium according to any one of claims 68 to 69, wherein the second diameter threshold includes a diameter range of approximately 50 micrometers (μm) to 100 μm.

71. The non-transitory computer-readable medium according to claim 65, wherein the instructions for determining whether at least one of the one or more diameter measurements meets the diameter threshold further include instructions for performing the following: For each highly reflective entity in the identified set of highly reflective entities: Associate an ellipse with the identified highly reflective entity; Determine the diameter of the longest axis of the ellipse; and Estimate at least one of the one or more diameter measurements based on the diameter of the longest axis of the ellipse.

72. The non-transitory computer-readable medium according to claim 65, wherein the instructions for identifying the HRFs in the retina of the eye of the patient further include instructions for classifying the eye of the patient as having at least one of the following: diabetic retinopathy (DR), diabetic macular edema (DME), age-related macular degeneration (AMD), neovascular age-related macular degeneration (nAMD), geographic atrophy (GA), macular atrophy (MA), or retinal vein occlusion (RVO).

73. The non-transitory computer-readable medium according to claim 65, wherein the instructions for identifying the HRFs in the retina of the eye of the patient further include instructions for performing the following: Access ETDRS grid mapping information that identifies one or more partitions of an Early Treatment Diabetic Retinopathy Study (ETDRS) grid; and Determine one or more volume measurements of the identified HRFs at least in part based on the ETDRS grid mapping information.

74. The non-transitory computer-readable medium according to claim 73, wherein the one or more volume measurement results include one or more of the following: the HRF volume in the retina of the eye of the patient, the HRF area in the retina of the eye of the patient, or the HRF thickness in the retina of the eye of the patient.

75. The non-transitory computer-readable medium according to any one of claims 73 to 74, wherein the instructions further include instructions for determining at least in part based on the ETDRS grid mapping information the HRF volume in the retina of the eye of the patient relative to at least one of the identified one or more partitions.

76. The non-transitory computer-readable medium according to claim 75, wherein the instructions for determining the HRF volume in the retina of the eye of the patient further include instructions for determining a decrease in the HRF volume in the retina of the eye of the patient.

77. The non-transitory computer-readable medium according to claim 75, wherein at least one of the identified one or more partitions includes the outer retina partition of the ETDRS grid.

78. The non-transitory computer-readable medium according to any one of claims 73 to 77, wherein the instructions further include instructions for determining at least in part based on the ETDRS grid mapping information the HRF count in the retina of the eye of the patient relative to at least one of the identified one or more partitions.

79. The non-transitory computer-readable medium according to any one of claims 73 to 78, wherein the instructions further include instructions for performing the following: accessing a frontal image of the retina of the eye of the patient, wherein the frontal image is associated with the one or more OCT scans; and mapping the identified HRF to the frontal image at least in part based on the ETDRS grid mapping information.

80. The non-transitory computer-readable medium according to claim 65, wherein the one or more machine learning models include at least one semantic segmentation model.

81. The non-transitory computer-readable medium according to claim 80, wherein the at least one semantic segmentation model includes a U-Net architecture.

82. The non-transitory computer-readable medium according to claim 65, wherein the instructions further include instructions for training the one or more machine learning models by: accessing an OCT scan data set of the retinas of the eyes of one or more patients, wherein the OCT scan data set includes sparse annotations of HRF in the retinas of the eyes of the one or more patients; dividing the OCT scan data set into a model training data set and a model validation data set; Training the one or more machine learning models based on the model training dataset to segment OCT scans to identify a set of highly reflective entities detectable from the OCT scans, wherein the identified set of highly reflective entities includes highly reflective material (HRM); and Evaluating the one or more machine learning models based on the model validation dataset.

83. The non-transitory computer-readable medium according to claim 82, wherein the instructions further include instructions for identifying HRF in the retina of the eyes of the one or more patients based on whether one or more diameter measurements of the HRM meet a predetermined diameter threshold.

84. The non-transitory computer-readable medium according to claim 82, wherein the sparse annotation of HRF includes boundary geometries surrounding multiple HRF instances, and wherein the instructions further include instructions for: Before training the one or more machine learning models, adjusting the boundary geometries by reducing the size of the boundary geometries, the size of the boundary geometries being reduced to annotate individual HRF instances.

85. The non-transitory computer-readable medium according to claim 65, wherein the one or more OCT scans include one or more first OCT scans of the retina of the eyes of the patient captured on an initial date, wherein the identified HRF includes a first HRF volume, and wherein the instructions further include instructions for: accessing one or more second OCT scans of the retina of the eyes of the patient; Inputting the one or more second OCT scans into the one or more machine learning models to segment the one or more second OCT scans to identify a second set of highly reflective entities detectable from the one or more second OCT scans; Based on the segmented one or more second OCT scans, determining one or more second diameter measurements corresponding to each highly reflective entity in the identified second set of highly reflective entities; Identifying a second HRF volume in the retina of the eyes of the patient based on whether at least one of the one or more second diameter measurements meets the diameter threshold; and Determining whether the eyes of the patient respond to treatment based on the second HRF volume.

86. The non-transitory computer-readable medium according to claim 85, wherein the instructions further include instructions for determining the degree to which the eyes of the patient respond to the treatment based on the second HRF volume.

87. The non-transitory computer-readable medium according to any one of claims 85 to 86, wherein the eyes of the patient respond to the treatment when the second HRF volume is less than the first HRF volume.

88. The non-transitory computer-readable medium according to any one of claims 85 to 87, wherein the treatment comprises an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-vascular endothelial growth factor-A (anti-VEGF-A) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof.

89. The non-transitory computer-readable medium according to claims 85 to 88, wherein the anti-VEGF-A antibody comprises faricimab.

90. The non-transitory computer-readable medium according to claims 85 to 89, wherein the anti-Ang-2 antibody comprises faricimab.

91. The non-transitory computer-readable medium according to claims 85 to 90, wherein the anti-VEGF antibody is selected from the group consisting of ranibizumab, aflibercept, brolucizumab, bevacizumab, and pegaptanib sodium.

92. The non-transitory computer-readable medium according to claim 65, wherein the instructions further comprise instructions for identifying an effective treatment regimen for an anti-vascular endothelial growth factor (anti-VEGF) antibody, an anti-angiopoietin-2 (anti-Ang-2) antibody, or a combination thereof to treat the patient's eye based on the volume or number of the identified HRF.

93. The non-transitory computer-readable medium according to claim 92, wherein the instructions for identifying the effective treatment regimen further comprise instructions for identifying a dose for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRF.

94. The non-transitory computer-readable medium according to any one of claims 92 to 93, wherein the instructions for identifying the effective treatment regimen further comprise instructions for identifying a schedule for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRF.

95. The non-transitory computer-readable medium according to any one of claims 92 to 94, wherein the instructions for identifying the effective treatment regimen further comprise instructions for identifying a duration for administering the anti-VEGF antibody, the anti-Ang-2 antibody, or a combination thereof based on the volume or the number of the identified HRF.

96. The non-transitory computer-readable medium according to claim 85, wherein the one or more second OCT scans are captured on one or more dates selected from the group consisting of approximately 0.25 months, 0.5 months, 0.75 months, 1 month, 3 months, 6 months, 9 months, 12 months, 15 months, 18 months, 21 months, 24 months, 27 months, 30 months, 33 months, or 36 months from the initial date.

97. The method according to any one of claims 1 to 32, further comprising: Determine whether at least one of the one or more diameter measurements meets a third diameter threshold.

98. The method according to claim 97, wherein the third diameter threshold includes a minimum diameter of approximately 100 micrometers (μm).

99. The system according to any one of claims 33 to 64, further comprising: Determine whether at least one of the one or more diameter measurements satisfies a third diameter threshold.

100. The system according to claim 99, wherein the third diameter threshold includes a minimum diameter of approximately 100 micrometers (μm).

101. The non-transitory computer-readable medium according to any one of claims 65 to 96, further comprising: Determine whether at least one of the one or more diameter measurements satisfies a third diameter threshold.

102. The non - transitory computer - readable medium according to claim 101, wherein the third diameter threshold includes a minimum diameter of approximately 100 micrometers (μm).

Images