{"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/an-ensemble-deep-learning-based-approach-for","title":"An Ensemble Deep Learning Based Approach for Red Lesion Detection in Fundus Images","arxiv_id":"1706.03008","date":"2017-06-09","proceeding":null,"authors":["José Ignacio Orlando","Elena Prokofyeva","Mariana del Fresno","Matthew B. Blaschko"],"abstract":"Diabetic retinopathy is one of the leading causes of preventable blindness in\nthe world. Its earliest sign are red lesions, a general term that groups both\nmicroaneurysms and hemorrhages. In daily clinical practice, these lesions are\nmanually detected by physicians using fundus photographs. However, this task is\ntedious and time consuming, and requires an intensive effort due to the small\nsize of the lesions and their lack of contrast. Computer-assisted diagnosis of\nDR based on red lesion detection is being actively explored due to its\nimprovement effects both in clinicians consistency and accuracy. Several\nmethods for detecting red lesions have been proposed in the literature, most of\nthem based on characterizing lesion candidates using hand crafted features, and\nclassifying them into true or false positive detections. Deep learning based\napproaches, by contrast, are scarce in this domain due to the high expense of\nannotating the lesions manually. In this paper we propose a novel method for\nred lesion detection based on combining both deep learned and domain knowledge.\nFeatures learned by a CNN are augmented by incorporating hand crafted features.\nSuch ensemble vector of descriptors is used afterwards to identify true lesion\ncandidates using a Random Forest classifier. We empirically observed that\ncombining both sources of information significantly improve results with\nrespect to using each approach separately. Furthermore, our method reported the\nhighest performance on a per-lesion basis on DIARETDB1 and e-ophtha, and for\nscreening and need for referral on MESSIDOR compared to a second human expert.\nResults highlight the fact that integrating manually engineered approaches with\ndeep learned features is relevant to improve results when the networks are\ntrained from lesion-level annotated data. An open source implementation of our\nsystem is publicly available online.","url_abs":"http://arxiv.org/abs/1706.03008v2","url_pdf":"http://arxiv.org/pdf/1706.03008v2.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":"an-ensemble-deep-learning-based-approach-for","repo_url":"https://github.com/ignaciorlando/red-lesion-detection","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"lesion-detection","task_name":"Lesion Detection"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}