[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-new-technique-helps-ai-segment-organs-from-incomplete-scan-labels":10,"sections":34},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":29,"feedback":33,"feedback_at":22,"cost_usd":33,"total_tokens":33},6828,"new-technique-helps-ai-segment-organs-from-incomplete-scan-labels","New Technique Helps AI Segment Organs From Incomplete Scan Labels","A two-stage AI framework uses optimal transport to align organ features so segmentation stays accurate despite missing labels.","A new AI framework learns to spot organs in medical scans even when most of the training labels are missing.\n\nResearchers built a two-stage system for multi-organ segmentation. In the first stage, the model trains on whatever annotations exist to build solid feature representations for each organ. In the second stage, it adds learnable \"organ prototypes\" and a Sinkhorn-triplet loss - a distance-based training signal drawn from optimal transport - to pull the same organ's features together across different datasets and push different organs' features apart, even in scans that were never labeled for those organs. Tested on the BTCV benchmark, the approach matched state-of-the-art segmentation accuracy while staying computationally efficient.\n\nMedical imaging datasets are notoriously incomplete - one hospital's scans might label the liver, another's the kidneys, and few label everything. Prior fixes leaned on pseudo-labels, essentially guessing at missing annotations and hoping the guesses held up; this method instead aligns the underlying feature space directly, a more principled way to handle the domain shift between imaging sources.\n\nIt is a research paper, not a hospital-ready product, but if the results hold up past one benchmark dataset, it could cut the annotation burden that keeps a lot of promising segmentation models stuck in the lab.","[\"ai\",\"medical-imaging\",\"computer-vision\",\"research\"]","2026-09-18T04:00:00.000Z","2026-09-18T18:29:47.143Z","2026-09-18T18:29:59.046Z","published",null,[],"ai",[24,26,27,28],"medical-imaging","computer-vision","research",[30],{"name":31,"url":32},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.19176",0,{"sections":35},[36,39,43,48,53,57,61,66,70,75,80,85,90,95],{"name":37,"slug":24,"count":38,"latest_published_at":18},"AI",4016,{"name":40,"slug":41,"count":42,"latest_published_at":18},"Security","security",653,{"name":44,"slug":45,"count":46,"latest_published_at":47},"Policy","policy",338,"2026-09-11T04:00:00.000Z",{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":18},"Hardware","hardware",155,{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",121,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",99,"2026-09-09T17:27:33.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":18},"Dev Tools","dev-tools",77,{"name":71,"slug":72,"count":73,"latest_published_at":74},"Software","software",75,"2026-09-10T20:41:21.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Startups","startups",55,"2026-09-09T23:14:29.000Z",{"name":81,"slug":82,"count":83,"latest_published_at":84},"Gaming","gaming",43,"2026-09-10T12:18:06.000Z",{"name":86,"slug":87,"count":88,"latest_published_at":89},"General","general",41,"2026-09-08T01:57:23.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]