Abstract
Deep learning–based nuclei analysis in histopathology images typically relies on large-scale expert-curated datasets, which are costly to construct and difficult to scale across diverse organs and staining conditions. To address this limitation, we propose a framework that leverages cell-level spatial transcriptomics (ST) as an alternative source of supervision for nuclei classification. Cell-level ST provides both gene expression profiles and their corresponding nuclei masks, enabling the generation of image-label pairs without manual cell-type and pixel-level nuclei annotation.
Specifically, we convert gene expression profiles into cell-type labels and use them to train image-based nuclei classifiers. Because existing cell-typing methods are designed for transcriptomic data rather than image recognition, we introduce an image-oriented classification strategy that effectively transfers gene expression–derived cell identities to histopathology images.
To evaluate generalization, we train models on a limited set of organs and evaluate them on previously unseen tissues. Experimental results demonstrate that our framework achieves superior segmentation and classification performance compared with conventional supervised approaches despite using fewer manually annotated datasets. These findings suggest that cell-level spatial transcriptomics can serve as a scalable source of supervision for pathology image analysis. Code will be released upon acceptance.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MISO_005.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=H0ichmpdgb
BibTex
@InProceedings{NisKaz_Leveraging_MICCAISAT2026,
author = { Nishimura, Kazuya AND Bise, Ryoma AND Hirose, Haruka AND Kojima, Yasuhiro},
title = { { Leveraging Cell-level Spatial Transcriptomics as Weak Supervision for Nuclei Classification } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17252},
month = {pending},
page = {pending}
}
