Detection-Guided ROI-Constrained Diffusion for Weakly Supervised White Blood Cell Segmentation: A Retrospective Internal and External Dataset Evaluation
Julius Bamwenda, Mehmet Siraç Özerdem, Orhan Ayyildiz, Veysi Akpolat, İrem Akpolat · 2026
Read Detection-Guided ROI-Constrained Diffusion for Weakly Supervised White Blood Cell Segmentation: A Retrospective Internal and External Dataset Evaluation by Julius Bamwenda, Mehmet Siraç Özerdem, Orhan Ayyildiz, Veysi Akpolat, İrem Akpolat (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/pmc-13566365.txt.
Description
Julius Bamwenda and co-authors build, in a 2026 study, a tool that outlines white blood cells in microscope images with little hand labeling. Pixel-by-pixel labels take much expert time, and models trained on one set can fail on the next. Their frame first finds each cell with a fast detector, then cuts a small region round it. A diffusion model, trained on rough auto-made masks, refines the outline in that region.
Expert masks were kept aside to score the true result. In a set of 14,721 records from Dicle, the rough masks scored about 0.70 on an overlap scale. That is usable but far from perfect, the team notes. Tests on the held-out part and on a second set of 11,200 records check how the frozen tool travels.
Catalog record
| CREATOR | Julius Bamwenda, Mehmet Siraç Özerdem, Orhan Ayyildiz, Veysi Akpolat, İrem Akpolat |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |