Development and validation of the Artificial Intelligence-Proliferative Vitreoretinopathy (AI-PVR) Insight system for deep learning-based diagnosis and postoperative risk prediction in proliferative vitreoretinopathy using multimodal fundus imaging

Abstract

Background: Early diagnosis of proliferative vitreoretinopathy (PVR) is crucial for preventing vision loss and ensuring effective treatment of retinal detachment. This study developed and validated a system named Artificial Intelligence-Proliferative Vitreoretinopathy (AI-PVR) Insight, which aims to automate the identification, grading, and postoperative risk assessment of PVR.

Methods: We retrospectively collected data from 1,700 eyes of 1,700 patients who underwent vitrectomy at the Eye Hospital affiliated with Nanchang University and at Jiangxi Provincial People's Hospital from June 2015 to December 2023 for the development and validation of the AI-PVR Insight system performance. This system is based on two deep learning models, TwinsSVT and DenseNet-121, which extract features from three modalities: B-scan ultrasound (B-scan), optical coherence tomography (OCT), and ultra-widefield (UWF) retinal imaging. After principal component analysis (PCA) dimension reduction and feature fusion, multi-layer perceptron (MLP) or support vector machine (SVM) classifiers are used to identify PVR, assess severity, and predict postoperative risks. The performance of the system was evaluated by calculating area under the curve (AUC) values, accuracy, precision, recall, and F1 scores.

Results: The AI-PVR Insight system demonstrated exceptional performance in PVR identification and severity grading, with AUC values exceeding 0.957 (0.902, 1.000) on internal and external test sets, respectively. For predicting postoperative PVR risk, the system achieved AUC values above 0.827 (0.737, 0.916) on both test sets, respectively.

Conclusions: The AI-PVR Insight system has successfully achieved automatic identification, grading, and postoperative risk assessment of PVR, providing clinical physicians with support in formulating more targeted treatment plans and delivering critical insights for the effective prevention and management of PVR progression.

Keywords: Deep learning; multicenter; multimodal; proliferative vitreoretinopathy (PVR).

Figure 1

Flowchart illustrating the primary methodology of the current study. (A) Automatic identification process of PVR. (B) Automated grading process of PVR. (C) Automated prediction process of postoperative PVR risk in retinal detachment patients. HR, high risk; LR, low risk; MLP, multilayer perceptron; PVR, proliferative vitreoretinopathy; SVM, support vector machine.

Figure 2

Flowchart of sample selection. HR, high risk; LR, low risk; PVR, proliferative vitreoretinopathy.

Figure 3

Predictive performance of models on PVR-related tasks. (A) ROC curves for PVR identification: comparing the accuracy of models across validation, internal, and external test sets. (B) The PVR grading AUC bar chart: comparing the AUC of different models on different datasets. (C) Multimodal model’s ROC for PVR grading: presents the multimodal model’s performance in PVR grading. (D) Risk assessment ROC curves: measuring predictive accuracy for PVR risk. Class 0 = Grade A; Class 1 = Grade B; Class 2 = Grade C. AUC, area under the curve; AUROC, area under the receiver operating characteristic curve; BUS, B-scan model; CI, confidence interval; MUTI, multimodal model; OCT, optical coherence tomography model; PVR, proliferative vitreoretinopathy; ROC, receiver operating characteristic; UWF, ultra-widefield model.

Figure 4

Performance comparison between ophthalmologists and AI-PVR Insight before and after AI assistance. The ROC curves of ophthalmologists and AI-PVR Insight in the validation set (A), internal test set (B), and external test set (C) before AI assistance. The ROC curves of ophthalmologists and AI-PVR Insight in the validation set (A1), internal test set (B1), and external test set (C1) after AI assistance. The accuracy line charts of ophthalmologists at different career stages in identifying PVR in the validation set (A2), internal test set (B2), and external test set (C2) before and after AI assistance. AI, artificial intelligence; AUC, area under the curve; CI, confidence interval; PVR, proliferative vitreoretinopathy; ROC, receiver operating characteristic.