Po-Kang Lin1,2,3, Yu-Hsien Chiu4,5, Chiu-Jung Huang4,6, Chien-Yao Wang5, Mei-Lien Pan5,7, Da-Wei Wang5, Hong-Yuan Mark Liao5, Yong-Sheng Chen8, Chieh-Hsiung Kuan9,10, Shih-Yen Lin8, Li-Fen Chen4,6. 1. National Taiwan University, Graduate Institute of Biomedical Electronics and Bioinformatics, Taipei, Taiwan. 2. National Yang Ming Chiao Tung University, School of Medicine, Department of Ophthalmology, Taipei, Taiwan. 3. Taipei Veterans General Hospital, Department of Ophthalmology, Taipei, Taiwan. 4. National Yang Ming Chiao Tung University, Institute of Brain Science, College of Medicine, Taipei, Taiwan. 5. Academia Sinica, Institute of Information Science, Taipei, Taiwan. 6. Taipei Veterans General Hospital, Integrated Brain Research Unit, Department of Medical Research, Taipei, Taiwan. 7. National Yang Ming Chiao Tung University, Information Technology Service Center, Taipei, Taiwan. 8. National Yang Ming Chiao Tung University, Department of Computer Science, Hsinchu, Taiwan. 9. National Taiwan University, Department of Electrical Engineering, Taipei, Taiwan. 10. National Taiwan University, Gaduate Institute of Biomedical Electronics and Bioinformatics, Taipei, Taiwan.
Abstract
Purpose: Retinopathy screening via digital imaging is promising for early detection and timely treatment, and tracking retinopathic abnormality over time can help to reveal the risk of disease progression. We developed an innovative physician-oriented artificial intelligence-facilitating diagnosis aid system for retinal diseases for screening multiple retinopathies and monitoring the regions of potential abnormality over time. Approach: Our dataset contains 4908 fundus images from 304 eyes with image-level annotations, including diabetic retinopathy, age-related macular degeneration, cellophane maculopathy, pathological myopia, and healthy control (HC). The screening model utilized a VGG-based feature extractor and multiple-binary convolutional neural network-based classifiers. Images in time series were aligned via affine transforms estimated through speeded-up robust features. Heatmaps of retinopathy were generated from the feature extractor using gradient-weighted class activation mapping++, and individual candidate retinopathy sites were identified from the heatmaps using clustering algorithm. Nested cross-validation with a train-to-test split of 80% to 20% was used to evaluate the performance of the screening model. Results: Our screening model achieved 99% accuracy, 93% sensitivity, and 97% specificity in discriminating between patients with retinopathy and HCs. For discriminating between types of retinopathy, our model achieved an averaged performance of 80% accuracy, 78% sensitivity, 94% specificity, 79% F1-score, and Cohen's kappa coefficient of 0.70. Moreover, visualization results were also shown to provide reasonable candidate sites of retinopathy. Conclusions: Our results demonstrated the capability of the proposed model for extracting diagnostic information of the abnormality and lesion locations, which allows clinicians to focus on patient-centered treatment and untangles the pathological plausibility hidden in deep learning models.
Purpose: Retinopathy screening via digital imaging is promising for early detection and timely treatment, and tracking retinopathic abnormality over time can help to reveal the risk of disease progression. We developed an innovative physician-oriented artificial intelligence-facilitating diagnosis aid system for retinal diseases for screening multiple retinopathies and monitoring the regions of potential abnormality over time. Approach: Our dataset contains 4908 fundus images from 304 eyes with image-level annotations, including diabetic retinopathy, age-related macular degeneration, cellophane maculopathy, pathological myopia, and healthy control (HC). The screening model utilized a VGG-based feature extractor and multiple-binary convolutional neural network-based classifiers. Images in time series were aligned via affine transforms estimated through speeded-up robust features. Heatmaps of retinopathy were generated from the feature extractor using gradient-weighted class activation mapping++, and individual candidate retinopathy sites were identified from the heatmaps using clustering algorithm. Nested cross-validation with a train-to-test split of 80% to 20% was used to evaluate the performance of the screening model. Results: Our screening model achieved 99% accuracy, 93% sensitivity, and 97% specificity in discriminating between patients with retinopathy and HCs. For discriminating between types of retinopathy, our model achieved an averaged performance of 80% accuracy, 78% sensitivity, 94% specificity, 79% F1-score, and Cohen's kappa coefficient of 0.70. Moreover, visualization results were also shown to provide reasonable candidate sites of retinopathy. Conclusions: Our results demonstrated the capability of the proposed model for extracting diagnostic information of the abnormality and lesion locations, which allows clinicians to focus on patient-centered treatment and untangles the pathological plausibility hidden in deep learning models.
Authors: Felix Grassmann; Judith Mengelkamp; Caroline Brandl; Sebastian Harsch; Martina E Zimmermann; Birgit Linkohr; Annette Peters; Iris M Heid; Christoph Palm; Bernhard H F Weber Journal: Ophthalmology Date: 2018-04-10 Impact factor: 12.079
Authors: Kumari Neelam; Chiu Ming Gemmy Cheung; Kyoko Ohno-Matsui; Timothy Y Y Lai; Tien Y Wong Journal: Prog Retin Eye Res Date: 2012-04-28 Impact factor: 21.198