Collaborative instance-level and bag-level multiple instance learning with label disambiguation for whole slide image analysis.
New AI method improves cancer detection in pathology slides by better handling noisy data, with code freely available for researchers to implement.
Tencent AI Lab proposes CIB-MIL, a multiple instance learning method that addresses the noisy pseudo-label problem in whole-slide image analysis by combining instance-level pseudo-label disambiguation with bag-level attention supervision, achieving state-of-the-art performance across cancer pathology tasks. The code is publicly available, enabling adoption by the computational pathology community.
What the study was
- Study design
- ai_method_validation
- Category
- Diagnostics
- Maturity
- Exploratory
- Journal
- Med Image Anal
Why it surfaced
Medical Image Analysis is a top-tier AI/medical imaging journal; SOTA performance on 5 pathology datasets; Tencent AI Lab has strong track record; WSI-based cancer diagnosis AI is a key application on the AI/ML watchlist.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.