Satish K Panda1, Haris Cheong1, Tin A Tun2, Sripad K Devella1, Vijayalakshmi Senthil3, Ramaswami Krishnadas3, Martin L Buist4, Shamira Perera5, Ching-Yu Cheng6, Tin Aung6, Alexandre H Thiéry7, Michaël J A Girard8. 1. From the Ophthalmic Engineering & Innovation Laboratory (OEIL), Singapore Eye Research Institute, Singapore National Eye Centre (S.K.P., H.C., S.K.D., M.J.A.G.); Department of Biomedical Engineering, National University of Singapore (S.K.P., H.C., S.K.D., M.L.B.). 2. Singapore Eye Research Institute, Singapore National Eye Centre, Singapore (T.A.T., S.P., C.-Y.C., T.A.). 3. Glaucoma Services, Aravind Eye Care Systems, Madurai, India (V.S., R.K.). 4. Department of Biomedical Engineering, National University of Singapore (S.K.P., H.C., S.K.D., M.L.B.). 5. Singapore Eye Research Institute, Singapore National Eye Centre, Singapore (T.A.T., S.P., C.-Y.C., T.A.); Duke-NUS Medical School (S.P., C.-Y.C., T.A., M.J.A.G.). 6. Singapore Eye Research Institute, Singapore National Eye Centre, Singapore (T.A.T., S.P., C.-Y.C., T.A.); Duke-NUS Medical School (S.P., C.-Y.C., T.A., M.J.A.G.); Department of Ophthalmology, Yong Loo Lin School of Medicine (C.-Y.C., T.A.), National University of Singapore, Singapore. 7. Department of Statistics and Applied Probability (A.H.T.). 8. From the Ophthalmic Engineering & Innovation Laboratory (OEIL), Singapore Eye Research Institute, Singapore National Eye Centre (S.K.P., H.C., S.K.D., M.J.A.G.); Duke-NUS Medical School (S.P., C.-Y.C., T.A., M.J.A.G.). Electronic address: mgirard@ophthalmic.engineering.
Abstract
PURPOSE: To develop a novel deep-learning approach that can describe the structural phenotype of the glaucomatous optic nerve head (ONH) and can be used as a robust glaucoma diagnosis tool. DESIGN: Retrospective, deep-learning approach diagnosis study. METHOD: We trained a deep-learning network to segment 3 neural-tissue and 4 connective-tissue layers of the ONH. The segmented optical coherence tomography images were then processed by a customized autoencoder network with an additional parallel branch for binary classification. The encoder part of the autoencoder reduced the segmented optical coherence tomography images into a low-dimensional latent space (LS), whereas the decoder and the classification branches reconstructed the images and classified them as glaucoma or nonglaucoma, respectively. We performed principal component analysis on the latent parameters and identified the principal components (PCs). Subsequently, the magnitude of each PC was altered in steps and reported how it impacted the morphology of the ONH. RESULTS: The image reconstruction quality and diagnostic accuracy increased with the size of the LS. With 54 parameters in the LS, the diagnostic accuracy was 92.0 ± 2.3% with a sensitivity of 90.0 ± 2.4% (at 95% specificity), and the corresponding Dice coefficient for the reconstructed images was 0.86 ± 0.04. By changing the magnitudes of PC in steps, we were able to reveal how the morphology of the ONH changes as one transitions from a "nonglaucoma" to a "glaucoma" condition. CONCLUSIONS: Our network was able to identify novel biomarkers of the ONH for glaucoma diagnosis. Specifically, the structural features identified by our algorithm were found to be related to clinical observations of glaucoma.
PURPOSE: To develop a novel deep-learning approach that can describe the structural phenotype of the glaucomatous optic nerve head (ONH) and can be used as a robust glaucoma diagnosis tool. DESIGN: Retrospective, deep-learning approach diagnosis study. METHOD: We trained a deep-learning network to segment 3 neural-tissue and 4 connective-tissue layers of the ONH. The segmented optical coherence tomography images were then processed by a customized autoencoder network with an additional parallel branch for binary classification. The encoder part of the autoencoder reduced the segmented optical coherence tomography images into a low-dimensional latent space (LS), whereas the decoder and the classification branches reconstructed the images and classified them as glaucoma or nonglaucoma, respectively. We performed principal component analysis on the latent parameters and identified the principal components (PCs). Subsequently, the magnitude of each PC was altered in steps and reported how it impacted the morphology of the ONH. RESULTS: The image reconstruction quality and diagnostic accuracy increased with the size of the LS. With 54 parameters in the LS, the diagnostic accuracy was 92.0 ± 2.3% with a sensitivity of 90.0 ± 2.4% (at 95% specificity), and the corresponding Dice coefficient for the reconstructed images was 0.86 ± 0.04. By changing the magnitudes of PC in steps, we were able to reveal how the morphology of the ONH changes as one transitions from a "nonglaucoma" to a "glaucoma" condition. CONCLUSIONS: Our network was able to identify novel biomarkers of the ONH for glaucoma diagnosis. Specifically, the structural features identified by our algorithm were found to be related to clinical observations of glaucoma.
Authors: Joshua Muller; David Alonso-Caneiro; Scott A Read; Stephen J Vincent; Michael J Collins Journal: Transl Vis Sci Technol Date: 2022-02-01 Impact factor: 3.283
Authors: Chhavi Saini; Lucy Q Shen; Louis R Pasquale; Michael V Boland; David S Friedman; Nazlee Zebardast; Mojtaba Fazli; Yangjiani Li; Mohammad Eslami; Tobias Elze; Mengyu Wang Journal: Ophthalmol Sci Date: 2022-04-20