{"id":{"repo_id":"buffalo","oai_identifier":"oai:ubir.buffalo.edu:10477/86702"},"canonical_url":"https://search.dev.ndltd.org/etd/buffalo/oai:ubir.buffalo.edu:10477/86702","repository":{"repo_id":"buffalo","name":"Buffalo","base_url":"https://ubir.buffalo.edu/oai/request"},"display":{"title":"Few-Shot Feature Space Learning for Congenital Retinal Diseases Recognition","abstract":"M.S.","abstract_html":"M.S.","abstract_has_math":false,"creators":["Mai, Siwei"],"institution":"State University of New York at Buffalo","degree_name":null,"degree_level":null,"degree_discipline":null,"degree_department":null,"school":null,"contributors":["Gao, Mingchen","Computer Science and Engineering"],"advisors":[],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-02-21T21:36:38Z","date_published":"2025-02-21T21:36:38Z","updated_at":"2026-07-27T19:05:34Z","subjects":["artificial intelligence","ophthalmology"],"languages":["eng"],"rights":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"http://hdl.handle.net/10477/86702","outbound_label":"Handle","outbound_source":"dc:identifier"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor","label":"Contributor","values":["Gao, Mingchen","Computer Science and Engineering"]},{"key":"dc:creator","label":"Author","values":["Mai, Siwei"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date","label":"Dc Date","values":["2025-02-21T21:36:38Z","2020"]},{"key":"dc:publisher","label":"Institution","values":["State University of New York at Buffalo"]},{"key":"dc:type","label":"Dc Type","values":["Text","Thesis"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["artificial intelligence","ophthalmology"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language","label":"Dc Language","values":["eng"]},{"key":"dc:rights","label":"Dc Rights","values":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier","label":"Identifier","values":["http://hdl.handle.net/10477/86702"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description","label":"Description","values":["M.S.","The goal of this project is to recognize a rare congenital retinal disease, Hereditary Macular Degeneration, based on Optical Coherence Tomography (OCT) images, whose main manifestation is the confusion and adhesion of the layers of the retina. The challenge of using machine learning models to recognize rare diseases comes from the limited number of collected data. To address this problem, we propose to learn a discriminative feature space for the OCT images, on which many classifiers can be applied for various tasks. We formulate this problem as a few-shot learning task as only very limited samples are available. The Siamese training strategy with the triplet loss is employed to maximize the inter-class distance and minimize the intra-class distance. OCT images also have large variations due to different capturing devices and angles. To alleviate such effect, a pipeline of preprocessing is first utilized for image alignment. Tissue images at different angles can be roughly corrected to a horizontal state for better feature representation. There are 58 samples collected in our dataset and half of them are diseased. Extensive experiments on our dataset demonstrate the effectiveness of the proposed approach.","**To request an accessible version of the file(s) associated with this item, contact library@buffalo.edu. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.**"]},{"key":"dc:format","label":"Dc Format","values":["application/pdf"]},{"key":"dc:title","label":"Title","values":["Few-Shot Feature Space Learning for Congenital Retinal Diseases Recognition"]}]}],"canonical_facts":{"dc:contributor":["Gao, Mingchen","Computer Science and Engineering"],"dc:creator":["Mai, Siwei"],"dc:date":["2025-02-21T21:36:38Z","2020"],"dc:description":["M.S.","The goal of this project is to recognize a rare congenital retinal disease, Hereditary Macular Degeneration, based on Optical Coherence Tomography (OCT) images, whose main manifestation is the confusion and adhesion of the layers of the retina. The challenge of using machine learning models to recognize rare diseases comes from the limited number of collected data. To address this problem, we propose to learn a discriminative feature space for the OCT images, on which many classifiers can be applied for various tasks. We formulate this problem as a few-shot learning task as only very limited samples are available. The Siamese training strategy with the triplet loss is employed to maximize the inter-class distance and minimize the intra-class distance. OCT images also have large variations due to different capturing devices and angles. To alleviate such effect, a pipeline of preprocessing is first utilized for image alignment. Tissue images at different angles can be roughly corrected to a horizontal state for better feature representation. There are 58 samples collected in our dataset and half of them are diseased. Extensive experiments on our dataset demonstrate the effectiveness of the proposed approach.","**To request an accessible version of the file(s) associated with this item, contact library@buffalo.edu. Please include the item's persistent URL [http://hdl.handle.net/. . .] in your request.**"],"dc:format":["application/pdf"],"dc:identifier":["http://hdl.handle.net/10477/86702"],"dc:language":["eng"],"dc:publisher":["State University of New York at Buffalo"],"dc:rights":["Users of works found in University at Buffalo Institutional Repository (UBIR) are responsible for identifying and contacting the copyright owner for permission to reuse. University at Buffalo Libraries do not manage rights for copyright-protected works and cannot assist with permissions.","Copyright retained by author."],"dc:subject":["artificial intelligence","ophthalmology"],"dc:title":["Few-Shot Feature Space Learning for Congenital Retinal Diseases Recognition"],"dc:type":["Text","Thesis"]},"updated_at":"2026-07-27T19:05:34Z"}