[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-a-smarter-way-to-catch-bad-labels-in-medical-ct-data":10,"sections":35},{"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":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},4369,"a-smarter-way-to-catch-bad-labels-in-medical-ct-data","A Smarter Way to Catch Bad Labels in Medical CT Data","Researchers propose a tool that finds annotation errors in vascular CT scans without retraining a model or hiring a second radiologist.","A new framework can flag mislabeled regions in vascular CT datasets before those errors corrupt a model's training.\n\nAnnotating medical scans is expensive, so most datasets get a single human label per image. That creates a quiet problem: no one knows where the labels are wrong. The standard fixes — having multiple raters label the same scan, or baking error-detection into model training — are either costly or opaque. Researchers have now published a decoupled approach that works on the annotation itself, independent of any network, by exploiting a geometric property of blood vessels: cross-sections taken along a vessel's centreline look similar to one another and to cross-sections from other subjects. When two anatomically similar patches disagree on their masks, that disagreement is treated as evidence of a bad label.\n\nThe practical payoff is auditability. Instead of a black-box signal buried inside training loss, the method produces an explicit image-mask pair for every flagged region — something a radiologist or data engineer can actually inspect. Aggregated scores generate scan-level quality maps that can weight training samples or flag scans for re-review. On a coronary CT dataset, the tool found that transverse and oblique vessels carry roughly 5 times the annotation error rate of axis-aligned structures, a systematic bias that would silently skew any model trained on the raw data.\n\nThe result matters because medical AI pipelines are only as reliable as the labels feeding them. Regulators and hospital systems increasingly want to know not just that a model performs well on a benchmark, but that its training data was clean — and auditable. This framework gives teams a way to answer that question without doubling their annotation budget.\n\nThe code is public, which is a good sign; whether it survives contact with messier real-world datasets beyond the coronary CT benchmark is the next question worth asking.","[\"medical-ai\",\"computer-vision\",\"data-quality\",\"annotation\"]","2026-07-08T04:00:00.000Z","2026-07-08T07:24:51.017Z","2026-07-08T07:24:54.036Z","published",null,[],"ai",[26,27,28,29],"medical-ai","computer-vision","data-quality","annotation",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2607.05965",0,{"sections":36},[37,41,46,51,56,61,66,71,76,81,86,90,95,100],{"name":38,"slug":24,"count":39,"latest_published_at":40},"AI",2590,"2026-07-16T04:00:00.000Z",{"name":42,"slug":43,"count":44,"latest_published_at":45},"Security","security",294,"2026-07-15T19:59:48.000Z",{"name":47,"slug":48,"count":49,"latest_published_at":50},"Deals","deals",179,"2026-06-29T20:02:07.000Z",{"name":52,"slug":53,"count":54,"latest_published_at":55},"Policy","policy",158,"2026-07-16T00:02:48.000Z",{"name":57,"slug":58,"count":59,"latest_published_at":60},"Hardware","hardware",122,"2026-07-14T19:46:26.000Z",{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",93,"2026-07-13T13:20:48.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Software","software",70,"2026-07-13T19:52:25.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Science","science",66,"2026-07-10T10:29:37.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Dev Tools","dev-tools",59,"2026-07-07T04:00:00.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Gaming","gaming",41,"2026-07-09T04:00:00.000Z",{"name":87,"slug":88,"count":84,"latest_published_at":89},"Startups","startups","2026-06-29T20:55:50.000Z",{"name":91,"slug":92,"count":93,"latest_published_at":94},"General","general",29,"2026-07-10T22:28:58.000Z",{"name":96,"slug":97,"count":98,"latest_published_at":99},"Reviews","reviews",20,"2026-06-24T12:00:01.000Z",{"name":101,"slug":102,"count":103,"latest_published_at":104},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]