{"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/retinal-oct-disease-classification-with","title":"Retinal OCT disease classification with variational autoencoder regularization","arxiv_id":"1904.00790","date":"2019-03-23","proceeding":null,"authors":["Max-Heinrich Laves","Sontje Ihler","Lüder A. Kahrs","Tobias Ortmaier"],"abstract":"According to the World Health Organization, 285 million people worldwide live\nwith visual impairment. The most commonly used imaging technique for diagnosis\nin ophthalmology is optical coherence tomography (OCT). However, analysis of\nretinal OCT requires trained ophthalmologists and time, making a comprehensive\nearly diagnosis unlikely. A recent study established a diagnostic tool based on\nconvolutional neural networks (CNN), which was trained on a large database of\nretinal OCT images. The performance of the tool in classifying retinal\nconditions was on par to that of trained medical experts. However, the training\nof these networks is based on an enormous amount of labeled data, which is\nexpensive and difficult to obtain. Therefore, this paper describes a method\nbased on variational autoencoder regularization that improves classification\nperformance when using a limited amount of labeled data. This work uses a\ntwo-path CNN model combining a classification network with an autoencoder (AE)\nfor regularization. The key idea behind this is to prevent overfitting when\nusing a limited training dataset size with small number of patients. Results\nshow superior classification performance compared to a pre-trained and fully\nfine-tuned baseline ResNet-34. Clustering of the latent space in relation to\nthe disease class is distinct. Neural networks for disease classification on\nOCTs can benefit from regularization using variational autoencoders when\ntrained with limited amount of patient data. Especially in the medical imaging\ndomain, data annotated by experts is expensive to obtain.","url_abs":"http://arxiv.org/abs/1904.00790v1","url_pdf":"http://arxiv.org/pdf/1904.00790v1.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":[{"paper_slug":"retinal-oct-disease-classification-with","repo_url":"https://github.com/mlaves/oct-classification-vae-regularization","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"clustering","task_name":"Clustering"},{"task_slug":"diagnostic","task_name":"Diagnostic"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"retinal-oct-disease-classification","task_name":"Retinal OCT Disease Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}