Explore the vast range of titles available.

The volume and complexity of diagnostic imaging is increasing at a pace faster than the availability of human expertise to interpret it. Artificial intelligence has shown great promise in classifying two-dimensional photographs of some common diseases and typically relies on databases of millions of annotated images. Until now, the challenge of reaching the performance of expert clinicians in a real-world clinical pathway with three-dimensional diagnostic scans has remained unsolved. Here, we apply a novel deep learning architecture to a clinically heterogeneous set of three-dimensional optical coherence tomography scans from patients referred to a major eye hospital. We demonstrate performance in making a referral recommendation that reaches or exceeds that of experts on a range of sight-threatening retinal diseases after training on only 14,884 scans. Moreover, we demonstrate that the tissue segmentations produced by our architecture act as a device-independent representation; referral accuracy is maintained when using tissue segmentations from a different type of device. Our work removes previous barriers to wider clinical use without prohibitive training data requirements across multiple pathologies in a real-world setting. A novel deep learning architecture performs device-independent tissue segmentation of clinical 3D retinal images followed by separate diagnostic classification that meets or exceeds human expert clinical diagnoses of retinal disease.

An engineered patch of retinal pigment epithelium generated from human embryonic stem cells is transplanted into the eyes of two patients. Age-related macular degeneration (AMD) remains a major cause of blindness, with dysfunction and loss of retinal pigment epithelium (RPE) central to disease progression. We engineered an RPE patch comprising a fully differentiated, human embryonic stem cell (hESC)–derived RPE monolayer on a coated, synthetic basement membrane. We delivered the patch, using a purpose-designed microsurgical tool, into the subretinal space of one eye in each of two patients with severe exudative AMD. Primary endpoints were incidence and severity of adverse events and proportion of subjects with improved best-corrected visual acuity of 15 letters or more. We report successful delivery and survival of the RPE patch by biomicroscopy and optical coherence tomography, and a visual acuity gain of 29 and 21 letters in the two patients, respectively, over 12 months. Only local immunosuppression was used long-term. We also present the preclinical surgical, cell safety and tumorigenicity studies leading to trial approval. This work supports the feasibility and safety of hESC-RPE patch transplantation as a regenerative strategy for AMD.

Despite advances in artificial intelligence (AI), its application in medical imaging has been burdened and limited by expert-generated labels. We used images from optical coherence tomography angiography (OCTA), a relatively new imaging modality that measures retinal blood flow, to train an AI algorithm to generate flow maps from standard optical coherence tomography (OCT) images, exceeding the ability and bypassing the need for expert labeling. Deep learning was able to infer flow from single structural OCT images with similar fidelity to OCTA and significantly better than expert clinicians (P < 0.00001). Our model allows generating flow maps from large volumes of previously collected OCT data in existing clinical trials and clinical practice. This finding demonstrates a novel application of AI to medical imaging, whereby subtle regularities between different modalities are used to image the same body part and AI is used to generate detailed inferences of tissue function from structure imaging.

A pilot cross sectional study was conducted to investigate the role of red blood cells (RBC) deformability in type 2 diabetes mellitus (T2DM) without and with diabetic retinopathy (DR) using a dual optical tweezers stretching technique. A dual optical tweezers was made by splitting and recombining a single Nd:YAG laser beam. RBCs were trapped directly (i.e., without microbead handles) in the dual optical tweezers where they were observed to adopt a “side-on” orientation. RBC initial and final lengths after stretching were measured by digital video microscopy and a Deformability index (DI) calculated. Blood from 8 healthy controls, 5 T2DM and 7 DR patients with respective mean age of 52.4yrs, 51.6 yrs and 52 yrs was analysed. Initial average length of RBCs for control group was 8.45 ± 0.25 μm, 8.68 ± 0.49 μm for DM RBCs and 8.82 ± 0.32 μm for DR RBCs (p < 0.001). The DI for control group was 0.0698 ± 0.0224 and that for DM RBCs was 0.0645 ± 0.03 and 0.0635 ± 0.028 (p < 0.001) for DR group. DI was inversely related to basal length of RBCs (p = 0.02). DI of RBC from DM and DR patients was significantly lower in comparison with normal healthy controls. A dual optical tweezers method can hence be reliably used to assess RBC deformability.

Purpose of ReviewThis paper systematically reviews the recent progress in diabetic retinopathy screening. It provides an integrated overview of the current state of knowledge of emerging techniques using artificial intelligence integration in national screening programs around the world. Existing methodological approaches and research insights are evaluated. An understanding of existing gaps and future directions is created.Recent FindingsOver the past decades, artificial intelligence has emerged into the scientific consciousness with breakthroughs that are sparking increasing interest among computer science and medical communities. Specifically, machine learning and deep learning (a subtype of machine learning) applications of artificial intelligence are spreading into areas that previously were thought to be only the purview of humans, and a number of applications in ophthalmology field have been explored. Multiple studies all around the world have demonstrated that such systems can behave on par with clinical experts with robust diagnostic performance in diabetic retinopathy diagnosis. However, only few tools have been evaluated in clinical prospective studies.SummaryGiven the rapid and impressive progress of artificial intelligence technologies, the implementation of deep learning systems into routinely practiced diabetic retinopathy screening could represent a cost-effective alternative to help reduce the incidence of preventable blindness around the world.

Spatially-resolved retinal function can be measured by psychophysical testing like fundus-controlled perimetry (FCP or ‘microperimetry’). It may serve as a performance outcome measure in emerging interventional clinical trials for macular diseases as requested by regulatory agencies. As FCP constitute laborious examinations, we have evaluated a machine-learning-based approach to predict spatially-resolved retinal function (’inferred sensitivity’) based on microstructural imaging (obtained by spectral domain optical coherence tomography) and patient data in recessive Stargardt disease. Using nested cross-validation, prediction accuracies of (mean absolute error, MAE [95% CI]) 4.74 dB [4.48–4.99] were achieved. After additional inclusion of limited FCP data, the latter reached 3.89 dB [3.67–4.10] comparable to the test–retest MAE estimate of 3.51 dB [3.11–3.91]. Analysis of the permutation importance revealed, that the IS&OS and RPE thickness were the most important features for the prediction of retinal sensitivity. ’Inferred sensitivity’, herein, enables to accurately estimate differential effects of retinal microstructure on spatially-resolved function in Stargardt disease, and might be used as quasi-functional surrogate marker for a refined and time-efficient investigation of possible functionally relevant treatment effects or disease progression.

Background/aimRecently, there has been an increasing clinical need for objective evaluation of corneal neovascularisation, a condition which cause significant ocular morbidity. We describe the use of a rapid, non-invasive ‘en face’ optical coherence tomography angiography (OCTA) system for the assessment of corneal neovascularisation.MethodsConsecutive patients with abnormal corneal neovascularisation were scanned using a commercially available AngioVue OCTA system (Optovue, Fremont, California, USA) with the split-spectrum amplitude decorrelation angiography algorithm, using an anterior segment lens adapter. Each subject had four scans in each eye by a trained operator and two independent masked assessors analysed all images. Main outcome measures were scan quality (signal strength, image quality), area of neovascularisation and repeatability of corneal vascular grade.ResultsWe performed OCTA in 20 patients (11 men, 9 women, mean age 49.27±17.23 years) with abnormal corneal neovascularisation. The mean area of corneal neovascularisation was 0.57±0.30 mm2 with a mean neovascularisation grade of 3.5±0.2 in the OCTA scans. We found the OCTA to produce good quality images of the corneal vessels (signal strength: 36.95±13.97; image quality score 2.72±1.07) with good repeatability for assessing neovascularisation grade (κ=0.84).ConclusionsIn this preliminary clinical study, we describe a method for acquiring angiography images with ‘en face’ views, using an OCTA system adapted for the evaluation of corneal neovascularisation. Further studies are required to compare the scans to other invasive angiography techniques for the quantitative evaluation of abnormal corneal vessels.

To benchmark the human and machine performance of spectral-domain (SD) and swept-source (SS) optical coherence tomography (OCT) image segmentation, i.e., pixel-wise classification, for the compartments vitreous, retina, choroid, sclera. A convolutional neural network (CNN) was trained on OCT B-scan images annotated by a senior ground truth expert retina specialist to segment the posterior eye compartments. Independent benchmark data sets (30 SDOCT and 30 SSOCT) were manually segmented by three classes of graders with varying levels of ophthalmic proficiencies. Nine graders contributed to benchmark an additional 60 images in three consecutive runs. Inter-human and intra-human class agreement was measured and compared to the CNN results. The CNN training data consisted of a total of 6210 manually segmented images derived from 2070 B-scans (1046 SDOCT and 1024 SSOCT; 630 C-Scans). The CNN segmentation revealed a high agreement with all grader groups. For all compartments and groups, the mean Intersection over Union (IOU) score of CNN compartmentalization versus group graders' compartmentalization was higher than the mean score for intra-grader group comparison. The proposed deep learning segmentation algorithm (CNN) for automated eye compartment segmentation in OCT B-scans (SDOCT and SSOCT) is on par with manual segmentations by human graders.

AimsTo discuss foveal development in the context of detailed retinal vasculature imaging in foveal hypoplasia using optical coherence tomography angiography.MethodsIn this case series, the optical coherence tomography angiography results of four patients with idiopathic foveal hypoplasia and two patients with foveal hypoplasia secondary to oculocutaneous albinism are presented.ResultsCases with intact visual acuity demonstrated lower grades of foveal hypoplasia on optical coherence tomography, while those with poor vision demonstrated high grades of foveal hypoplasia. The superficial retinal capillary plexus was intact in the foveal area in all cases, with no demonstrable foveal avascular zone. The deep retinal capillary plexus was absent to variable degrees in most cases, but was most persistent in those cases with reduced vision.ConclusionsThe superficial retinal capillary plexus is present in cases with foveal hypoplasia, while the deep retinal capillary plexus is absent to varying degrees. Our findings support the hypothesis that an intact foveal avascular zone of the deep capillary plexus allows for outer retinal photoreceptor specialisation to occur unimpeded, resulting in preserved visual acuity, while this process may be inhibited by an absent deep capillary foveal avascular zone with resultant poor vision.

Background/AimTo describe an optical coherence tomography angiography (OCTA) system adapted for anterior segment imaging, compared with indocyanine green angiography (ICGA) in eyes with corneal vascularisation.MethodsRetrospective study of subjects with corneal vascularisation secondary to microbial keratitis who had OCTA scans performed using a commercially available split-spectrum amplitude-decorrelation algorithm angiography system (AngioVue; Optovue Inc., Fremont, California, USA) and ICGA images (Spectralis; Heidelberg Engineering, Heidelberg, Germany). The agreement between OCTA and ICGA techniques in terms of area of vascularisation measured, using Bland–Altman 95% limits of agreement (LOA).ResultsWe compared the area of corneal vascularisation in 64 scan images (eight eyes, four scans for each angiography technique). In our series, the overall mean area of vascularisation from the ICGA scans was 0.49±0.34 mm2 and OCTA scans was 0.51±0.36 mm2. We obtained substantial repeatability in terms of image quality score (κ=0.80) for all OCTA scans. The agreement between OCTA and ICGA scans was good, although ICGA measured a smaller area compared with the OCTA with a mean difference of −0.03 mm2 (95% CI −0.07 to 0.01). The LOA ranged from a lower limit of −0.27 (95% CI −0.34 to −0.19) to an upper limit of 0.20 (95% CI 0.13 to 0.28, p=0.127).ConclusionsWe found that rapid, non-contact OCTA adapted for the cornea was comparable with ICGA for measurement of the area of corneal vascularisation in this pilot clinical study. Further prospective studies are required to confirm if this relatively new imaging technique may be further developed to replace invasive angiography techniques for the anterior segment.