Abstract:The purpose of this study was to detect the presence of retinitis pigmentosa (RP) based on color fundus photographs using a deep learning model. A total of 1670 color fundus photographs from the Taiwan inherited retinal degeneration project and National Taiwan University Hospital were acquired and preprocessed. The fundus photographs were labeled RP or normal and divided into training and validation datasets (n = 1284) and a test dataset (n = 386). Three transfer learning models based on pre-trained Inception V3, Inception Resnet V2, and Xception deep learning architectures, respectively, were developed to classify the presence of RP on fundus images. The model sensitivity, specificity, and area under the receiver operating characteristic (AUROC) curve were compared. The results from the best transfer learning model were compared with the reading results of two general ophthalmologists, one retinal specialist, and one specialist in retina and inherited retinal degenerations. A total of 935 RP and 324 normal images were used to train the models. The test dataset consisted of 193 RP and 193 normal images. Among the three transfer learning models evaluated, the Xception model had the best performance, achieving an AUROC of 96.74%. Gradient-weighted class activation mapping indicated that the contrast between the periphery and the macula on fundus photographs was an important feature in detecting RP. False-positive results were mostly obtained in cases of high myopia with highly tessellated retina, and false-negative results were mostly obtained in cases of unclear media, such as cataract, that led to a decrease in the contrast between the peripheral retina and the macula. Our model demonstrated the highest accuracy of 96.00%, which was comparable with the average results of 81.50%, of the other four ophthalmologists. Moreover, the accuracy was obtained at the same level of sensitivity (95.71%), as compared to an inherited retinal disease specialist. RP is an important disease, but its early and precise diagnosis is challenging. We developed and evaluated a transfer-learning-based model to detect RP from color fundus photographs. The results of this study validate the utility of deep learning in automating the identification of RP from fundus photographs.

Weakly-Supervised Lesion-Aware and Consistency Regularization for Retinitis Pigmentosa Detection from Ultra-Widefield Images

Adapting the Segment Anything Model for Multi-modal Retinal Anomaly Detection and Localization

Weakly Supervised Lesion Detection from Fundus Images

A Detectability Analysis of Retinitis Pigmetosa Using Novel SE-ResNet Based Deep Learning Model and Color Fundus Images

Artificial Intelligence–Assisted Early Detection of Retinitis Pigmentosa — the Most Common Inherited Retinal Degeneration

UAU-Net: United Attention U-Shaped Network for the Segmentation of Pigment Deposits in Fundus Images of Retinitis Pigmentosa

Multi-input Deep Learning Model for RP Diagnosis Using FVEP and Prior Knowledge

Rotation-Oriented Collaborative Self-Supervised Learning for Retinal Disease Diagnosis

Detecting multiple retinal diseases in ultra-widefield fundus imaging and data-driven identification of informative regions with deep learning

Artificial intelligence-assisted management of retinal detachment from ultra-widefield fundus images based on weakly-supervised approach

Deep Multiple Instance Learning with Spatial Attention for ROP Case Classification, Instance Selection and Abnormality Localization.

Detecting Diverse Retinal Lesions Based on Low-Rank-Plus-Sparse Factorization over Normal Retinal Images

Multiscale triplet spatial information fusion-based deep learning method to detect retinal pigment signs with fundus images

Unsupervised Retinal Lesion Detection by Learning to Restore Corrupted Fundus Images

Automatic retinopathy of prematurity stagingwith a semi-supervised deep learning method based on dual consistency regularization

Pseudo-Labeling for Small Lesion Detection on Diabetic Retinopathy Images

Development and Validation of a Semi-Supervised Deep Learning Model for Automatic Retinopathy of Prematurity Staging

Regression and Learning with Pixel-wise Attention for Retinal Fundus Glaucoma Segmentation and Detection

Efficacy and accuracy of artificial intelligence to overlay multimodal images from different optical instruments in patients with retinitis pigmentosa

A Cross-Domain Weakly Supervised Diabetic Retinopathy Lesion Identification Method Based on Multiple Instance Learning and Domain Adaptation.

Intelligent Diagnosis of Multiple Peripheral Retinal Lesions in Ultra-widefield Fundus Images Based on Deep Learning