Abstract
Histopathological grading from hematoxylin and eosin (H&E)
stained whole-slide images (WSIs) is often trained as weakly supervised
multi-class classification, although diagnostic grades are ordered and the
zero/reference grade can play a distinct role relative to higher grades.We
introduce OrdPath-MIL, a case-level multiple-instance learning (MIL)
framework for ordinal WSI grading diagnosis. OrdPath-MIL combines
target-gated evidence aggregation, ordinal distribution modeling, and
negative evidence modeling to align patch evidence with grade ordering
and class-0-versus-higher-grade separation without patch-level labels.
We evaluate OrdPath-MIL on three H&E WSI grading cohorts with representative
MIL baselines, ablations, and heatmap visualization. Current
results show competitive ordinal grading performance and datasetdependent
benefits from the proposed ordinal diagnosis-aware modules.
Code: https://github.com/DigitalAlchemist-S/OrdPathMIL.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/COMPAYL_077.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
BibTex
@InProceedings{XuShi_OrdPathMIL_MICCAISAT2026,
author = { Xu, Shipu AND Ren, Ge AND Yeung, Ho Yin AND Cai, Yin AND Feng, Dingran AND Wong, Tak-siu AND Tang, Wai-Lun Victor AND Chan, Cheong Kin Ronald AND Chan, Angela Zaneta},
title = { { OrdPath-MIL: Ordinal Multiple Instance Learning for Pathology Whole-Slide Image Grading } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17251},
month = {pending},
page = {pending}
}