GDT-SwinKid: A hybrid model for precise renal lesion analysis
Thirupathi Rao N, V V Ramana CH, Eatedal Alabdulkreem, Ayman Aljarbouh, Samih M Mostafa · 2026
Read GDT-SwinKid: A hybrid model for precise renal lesion analysis by Thirupathi Rao N, V V Ramana CH, Eatedal Alabdulkreem, Ayman Aljarbouh, Samih M Mostafa (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0349285.txt.
Description
Thirupathi Rao N and colleagues built a hybrid deep learning model for finding and outlining kidney lesions in CT scans. Kidney disease varies widely and lesion edges can be faint, so radiologists face a slow, error-prone task. Their model, GDT-SwinKid, marries a transformer that grasps the whole scan's context with a decoder that restores fine detail, tuned by a statistical distribution step.
On clinical datasets, the model outlined lesions with overlap scores up to the mid-ninety percent range. It sorted cases with near-perfect discrimination, beating leading transformer and convolutional rivals by several points on overlap. Accurate kidney lesion maps support surgical planning and treatment tracking, so gains of this size are clinically meaningful if they hold in new hospitals.
Catalog record
| CREATOR | Thirupathi Rao N, V V Ramana CH, Eatedal Alabdulkreem, Ayman Aljarbouh, Samih M Mostafa |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 73988 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |