Abstract
The rapid growth of digital pathology has produced vast
repositories of hematoxylin and eosin (H&E) stained whole slide images.
However, many of these archives lack structured indexing and reliable
metadata, making systematic organization and retrieval challenging. Reverse
image search, also known as content-based image retrieval, addresses
this limitation by retrieving slides based on learned visual representations
rather than incomplete or inconsistent metadata. Although
retrieval systems have been deployed in digital pathology, many rely
on manually engineered indexing strategies or heuristic filtering rules,
which limit scalability and generalization across diverse diagnostic categories.
Thus, we propose CLEAR-WSI (Constant Length Embedding
& Automatic Retrieval), a fully automated slide-level retrieval framework.
CLEAR-WSI learns semantically structured whole slide embeddings
through an Attention-based Multiple Instance Learning framework,
compressing each WSI into a fixed dimensional representation for
efficient storage and scalable similarity search. Furthermore, we introduce
a self-reviewing diagnostic filtering mechanism that enforces label
consistency among retrieved candidates, improving diagnosis alignment
without relying on manually defined class-specific rules. Across two
public datasets, CAMELYON16 (lymph node metastases) and BRACS
(breast cancer subtypes), our diagnostic-aware method establishes new
state-of-the-art results, improving AccMV @5 from 77.49% to 89.92%
on CAMELYON16, from 54.12% to 75.86% on BRACS level-1, and
from 36.47% to 51.72% on BRACS level-2. Our annotation-free, datasetagnostic
search engine that scales across diverse data sources is openly
available: github.com/youssefwally/CLEAR-WSI
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/COMPAYL_038.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
BibTex
@InProceedings{WalYou_CLEARWSI_MICCAISAT2026,
author = { Wally, Youssef AND Liu, Jingsong AND Li, Han AND Dai, Jing AND Wetzer, Elisabeth AND Schüffler, Peter J.},
title = { { CLEAR-WSI: Towards Foundation-Model-Empowered Diagnosis Aligned Whole Slide Image Retrieval } },
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}
}