Search

Abstract Introduction Generative pretrained transformer-4 (GPT-4) has gained widespread attention from society, and its potential has been extensively evaluated in many areas. However, investigation of GPT-4’s use in medicine, especially in the ophthalmology field, is still limited. This study aims to evaluate GPT-4’s capability to identify rare ophthalmic diseases in three simulated scenarios for different end-users, including patients, family physicians, and junior ophthalmologists. Methods We selected ten treatable rare ophthalmic disease cases from the publicly available EyeRounds service. We gradually increased the amount of information fed into GPT-4 to simulate the scenarios of patient, family physician, and junior ophthalmologist using GPT-4. GPT-4’s responses were evaluated from two aspects: suitability (appropriate or inappropriate) and accuracy (right or wrong) by senior ophthalmologists (> 10 years’ experiences). Results Among the 30 responses, 83.3% were considered "appropriate" by senior ophthalmologists. In the scenarios of simulated patient, family physician, and junior ophthalmologist, seven (70%), ten (100%), and eight (80%) responses were graded as “appropriate” by senior ophthalmologists. However, compared to the ground truth, GPT-4 could only output several possible diseases generally without “right” responses in the simulated patient scenarios. In contrast, in the simulated family physician scenario, 50% of GPT-4's responses were “right,” and in the simulated junior ophthalmologist scenario, the model achieved a higher “right” rate of 90%. Conclusion To our knowledge, this is the first proof-of-concept study that evaluates GPT-4’s capacity to identify rare eye diseases in simulated scenarios involving patients, family physicians, and junior ophthalmologists. The results indicate that GPT-4 has the potential to serve as a consultation assisting tool for patients and family physicians to receive referral suggestions and an assisting tool for junior ophthalmologists to diagnose rare eye diseases. However, it is important to approach GPT-4 with caution and acknowledge the need for verification and careful referrals in clinical settings.

PurposeWe aim to develop a multi-task three-dimensional (3D) deep learning (DL) model to detect glaucomatous optic neuropathy (GON) and myopic features (MF) simultaneously from spectral-domain optical coherence tomography (SDOCT) volumetric scans.MethodsEach volumetric scan was labelled as GON according to the criteria of retinal nerve fibre layer (RNFL) thinning, with a structural defect that correlated in position with the visual field defect (i.e., reference standard). MF were graded by the SDOCT en face images, defined as presence of peripapillary atrophy (PPA), optic disc tilting, or fundus tessellation. The multi-task DL model was developed by ResNet with output of Yes/No GON and Yes/No MF. SDOCT scans were collected in a tertiary eye hospital (Hong Kong SAR, China) for training (80%), tuning (10%), and internal validation (10%). External testing was performed on five independent datasets from eye centres in Hong Kong, the United States, and Singapore, respectively. For GON detection, we compared the model to the average RNFL thickness measurement generated from the SDOCT device. To investigate whether MF can affect the model’s performance on GON detection, we conducted subgroup analyses in groups stratified by Yes/No MF. The area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, and accuracy were reported.ResultsA total of 8,151 SDOCT volumetric scans from 3,609 eyes were collected. For detecting GON, in the internal validation, the proposed 3D model had significantly higher AUROC (0.949 vs. 0.913, p < 0.001) than average RNFL thickness in discriminating GON from normal. In the external testing, the two approaches had comparable performance. In the subgroup analysis, the multi-task DL model performed significantly better in the group of “no MF” (0.883 vs. 0.965, p-value < 0.001) in one external testing dataset, but no significant difference in internal validation and other external testing datasets. The multi-task DL model’s performance to detect MF was also generalizable in all datasets, with the AUROC values ranging from 0.855 to 0.896.ConclusionThe proposed multi-task 3D DL model demonstrated high generalizability in all the datasets and the presence of MF did not affect the accuracy of GON detection generally.

Optical coherence tomography angiography (OCT-A) provides depth-resolved visualization of the retinal microvasculature without intravenous dye injection. It facilitates investigations of various retinal vascular diseases and glaucoma by assessment of qualitative and quantitative microvascular changes in the different retinal layers and radial peripapillary layer non-invasively, individually, and efficiently. Deep learning (DL), a subset of artificial intelligence (AI) based on deep neural networks, has been applied in OCT-A image analysis in recent years and achieved good performance for different tasks, such as image quality control, segmentation, and classification. DL technologies have further facilitated the potential implementation of OCT-A in eye clinics in an automated and efficient manner and enhanced its clinical values for detecting and evaluating various vascular retinopathies. Nevertheless, the deployment of this combination in real-world clinics is still in the “proof-of-concept” stage due to several limitations, such as small training sample size, lack of standardized data preprocessing, insufficient testing in external datasets, and absence of standardized results interpretation. In this review, we introduce the existing applications of DL in OCT-A, summarize the potential challenges of the clinical deployment, and discuss future research directions.

Advances in artificial intelligence deep learning (DL) have made tremendous impacts on the field of ocular imaging over the last few years. Specifically, DL has been utilised to detect and classify various ocular diseases on retinal photographs, optical coherence tomography (OCT) images, and OCT-angiography images. In order to achieve good robustness and generalisability of model performance, DL training strategies traditionally require extensive and diverse training datasets from various sites to be transferred and pooled into a “centralised location”. However, such a data transferring process could raise practical concerns related to data security and patient privacy. Federated learning (FL) is a distributed collaborative learning paradigm which enables the coordination of multiple collaborators without the need for sharing confidential data. This distributed training approach has great potential to ensure data privacy among different institutions and reduce the potential risk of data leakage from data pooling or centralisation. This review article aims to introduce the concept of FL, provide current evidence of FL in ocular imaging, and discuss potential challenges as well as future applications.

Summary: Background: Spectral-domain optical coherence tomography (SDOCT) can be used to detect glaucomatous optic neuropathy, but human expertise in interpretation of SDOCT is limited. We aimed to develop and validate a three-dimensional (3D) deep-learning system using SDOCT volumes to detect glaucomatous optic neuropathy. Methods: We retrospectively collected a dataset including 4877 SDOCT volumes of optic disc cube for training (60%), testing (20%), and primary validation (20%) from electronic medical and research records at the Chinese University of Hong Kong Eye Centre (Hong Kong, China) and the Hong Kong Eye Hospital (Hong Kong, China). Residual network was used to build the 3D deep-learning system. Three independent datasets (two from Hong Kong and one from Stanford, CA, USA), including 546, 267, and 1231 SDOCT volumes, respectively, were used for external validation of the deep-learning system. Volumes were labelled as having or not having glaucomatous optic neuropathy according to the criteria of retinal nerve fibre layer thinning on reliable SDOCT images with position-correlated visual field defect. Heatmaps were generated for qualitative assessments. Findings: 6921 SDOCT volumes from 1 384 200 two-dimensional cross-sectional scans were studied. The 3D deep-learning system had an area under the receiver operation characteristics curve (AUROC) of 0·969 (95% CI 0·960–0·976), sensitivity of 89% (95% CI 83–93), specificity of 96% (92–99), and accuracy of 91% (89–93) in the primary validation, outperforming a two-dimensional deep-learning system that was trained on en face fundus images (AUROC 0·921 [0·905–0·937]; p