Object knowledge-aware multiple instance learning for small tumor segmentation

doi:10.1016/j.bspc.2025.109400

Object knowledge-aware multiple instance learning for small tumor segmentation

2025-12-15

0

PRE

AI

DOI：10.1016/j.bspc.2025.109400

Original

Original request for help

Save

Abstract

En 中文

• Using bounding box annotations to train segmentation model can save time and cost. • Loose bounding boxes lack the tightness prior required for traditional MIL methods. • The accurate of OKMIL is approach to that of fully supervised segmentation methods. • The accurate of tiny tumor was improved by 14.85% compared to SOTA methods.

Object knowledge-aware multiple instance learning for small tumor segmentation

Abstract

Journal

Organization