{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/deep-learning-is-effective-for-the","title":"Deep learning is effective for the classification of OCT images of normal versus Age-related Macular Degeneration","arxiv_id":"1612.04891","date":"2016-12-15","proceeding":null,"authors":["Cecilia S. Lee","Doug M. Baughman","Aaron Y. Lee"],"abstract":"Objective: The advent of Electronic Medical Records (EMR) with large\nelectronic imaging databases along with advances in deep neural networks with\nmachine learning has provided a unique opportunity to achieve milestones in\nautomated image analysis. Optical coherence tomography (OCT) is the most\ncommonly obtained imaging modality in ophthalmology and represents a dense and\nrich dataset when combined with labels derived from the EMR. We sought to\ndetermine if deep learning could be utilized to distinguish normal OCT images\nfrom images from patients with Age-related Macular Degeneration (AMD). Methods:\nAutomated extraction of an OCT imaging database was performed and linked to\nclinical endpoints from the EMR. OCT macula scans were obtained by Heidelberg\nSpectralis, and each OCT scan was linked to EMR clinical endpoints extracted\nfrom EPIC. The central 11 images were selected from each OCT scan of two\ncohorts of patients: normal and AMD. Cross-validation was performed using a\nrandom subset of patients. Area under receiver operator curves (auROC) were\nconstructed at an independent image level, macular OCT level, and patient\nlevel. Results: Of an extraction of 2.6 million OCT images linked to clinical\ndatapoints from the EMR, 52,690 normal and 48,312 AMD macular OCT images were\nselected. A deep neural network was trained to categorize images as either\nnormal or AMD. At the image level, we achieved an auROC of 92.78% with an\naccuracy of 87.63%. At the macula level, we achieved an auROC of 93.83% with an\naccuracy of 88.98%. At a patient level, we achieved an auROC of 97.45% with an\naccuracy of 93.45%. Peak sensitivity and specificity with optimal cutoffs were\n92.64% and 93.69% respectively. Conclusions: Deep learning techniques are\neffective for classifying OCT images. These findings have important\nimplications in utilizing OCT in automated screening and computer aided\ndiagnosis tools.","url_abs":"http://arxiv.org/abs/1612.04891v1","url_pdf":"http://arxiv.org/pdf/1612.04891v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"retinal-oct-disease-classification","task_name":"Retinal OCT Disease Classification"},{"task_slug":"specificity","task_name":"Specificity"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/retinal-oct-disease-classification-on","task":"Retinal OCT Disease Classification","dataset":"Srinivasan2014","model":"Lee et al.","rank_in_archive_order":13,"of":14,"metrics":{"Acc":"87.63"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}