Maloca PM, Lee AY, de Carvalho ER, Okada M, Fasler K, Leung I, Hörmann B, Kaiser P, Suter S, Hasler PW, Zarranz-Ventura J, Egan C, Heeren TFC, Balaskas K, Tufail A, and Scholl HPN
Purpose: To benchmark the human and machine performance of spectral-domain (SD) and swept-source (SS) optical coherence tomography (OCT) image segmentation, i.e., pixel-wise classification, for the compartments vitreous, retina, choroid, sclera., Methods: A convolutional neural network (CNN) was trained on OCT B-scan images annotated by a senior ground truth expert retina specialist to segment the posterior eye compartments. Independent benchmark data sets (30 SDOCT and 30 SSOCT) were manually segmented by three classes of graders with varying levels of ophthalmic proficiencies. Nine graders contributed to benchmark an additional 60 images in three consecutive runs. Inter-human and intra-human class agreement was measured and compared to the CNN results., Results: The CNN training data consisted of a total of 6210 manually segmented images derived from 2070 B-scans (1046 SDOCT and 1024 SSOCT; 630 C-Scans). The CNN segmentation revealed a high agreement with all grader groups. For all compartments and groups, the mean Intersection over Union (IOU) score of CNN compartmentalization versus group graders' compartmentalization was higher than the mean score for intra-grader group comparison., Conclusion: The proposed deep learning segmentation algorithm (CNN) for automated eye compartment segmentation in OCT B-scans (SDOCT and SSOCT) is on par with manual segmentations by human graders., Competing Interests: Authors BH, PK, SS are salaried employees of Supercomputing Systems, Zurich; this does not alter our adherence to PLOS ONE policies on sharing data and materials. Outside of the present study, the authors declare the following competing interests: PMM is a consultant at Zeiss Forum, Roche and holds intellectual properties for machine learning at MIMO AG, Berne, Switzerland. AYL has received funding from Novartis, Microsoft Corporation, NVIDIA Corporation and grant number from NEI: K23EY029246. CE and AT received a financial grant from the National Institute for Health Research (NIHR) Biomedical Research Centre, based at Moorfields Eye Hospital, and also from the NHS Foundation Trust and the UCL Institute of Ophthalmology. The views expressed in this article are those of the authors and not necessarily those of the National Eye Institute, NHS, the NIHR, or the Department of Health. AT is a consultant for Heidelberg Engineering and Optovue and has received research grant funding from Novartis and Bayer. CE is a consultant for Heidelberg Engineering and has received research grant funding from Novartis. MO has received travel and honorarium from Allergan. KF has received fellowship support from Alfred Vogt Stipendium and Schweizerischer Fonds zur Verhütung und Bekämpfung der Blindheit and has been an external consultant for DeepMind. JZ-V declares the following (where C: Consultant, S: Speaker; TG: Travel Grant, G: Research Grant, IP: Intellectual Properties): Alcon (C,S, TG, Alimera Sciences (C, S, TG), Allergan (C, S, TG, G), Bausch & Lomb (S, TG), Bayer (C,S, TG), Brill Pharma (C, S9, Novartis (S, TG), Topcon (S, TG, Zeiss (S). These do not alter our adherence to PLOS ONE policies on sharing data and materials.