GrapHist: Graph Self-Supervised Learning for Histopathology

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Öğüt, Sevda, Vincent-Cuaz, Cédric, Dubljevic, Natalia, Hurtado, Carlos, Subramanian, Vaishnavi, Frossard, Pascal, Thanou, Dorina
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866908856885444608
author Öğüt, Sevda
Vincent-Cuaz, Cédric
Dubljevic, Natalia
Hurtado, Carlos
Subramanian, Vaishnavi
Frossard, Pascal
Thanou, Dorina
author_facet Öğüt, Sevda
Vincent-Cuaz, Cédric
Dubljevic, Natalia
Hurtado, Carlos
Subramanian, Vaishnavi
Frossard, Pascal
Thanou, Dorina
contents Self-supervised vision models have achieved notable success in digital pathology. However, their domain-agnostic transformer architectures are not originally designed to account for fundamental biological elements of histopathology images, namely cells and their complex interactions. In this work, we hypothesize that a biologically-informed modeling of tissues as cell graphs offers a more efficient representation learning. Thus, we introduce GrapHist, a novel graph-based self-supervised learning framework for histopathology, which learns generalizable and structurally-informed embeddings that enable diverse downstream tasks. GrapHist integrates masked autoencoders and heterophilic graph neural networks that are explicitly designed to capture the heterogeneity of tumor microenvironments. We pre-train GrapHist on a large collection of 11 million cell graphs derived from breast tissues and evaluate its transferability across in- and out-of-domain benchmarks. Our results show that GrapHist achieves competitive performance compared to its vision-based counterparts in slide-, region-, and cell-level tasks, while requiring four times fewer parameters. It also drastically outperforms fully-supervised graph models on cancer subtyping tasks. Finally, we also release five graph-based digital pathology datasets used in our study at https://huggingface.co/ogutsevda/datasets , establishing the first large-scale graph benchmark in this field. Our code is available at https://github.com/ogutsevda/graphist .
format Preprint
id arxiv_https___arxiv_org_abs_2603_00143
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GrapHist: Graph Self-Supervised Learning for Histopathology
Öğüt, Sevda
Vincent-Cuaz, Cédric
Dubljevic, Natalia
Hurtado, Carlos
Subramanian, Vaishnavi
Frossard, Pascal
Thanou, Dorina
Computer Vision and Pattern Recognition
Machine Learning
Self-supervised vision models have achieved notable success in digital pathology. However, their domain-agnostic transformer architectures are not originally designed to account for fundamental biological elements of histopathology images, namely cells and their complex interactions. In this work, we hypothesize that a biologically-informed modeling of tissues as cell graphs offers a more efficient representation learning. Thus, we introduce GrapHist, a novel graph-based self-supervised learning framework for histopathology, which learns generalizable and structurally-informed embeddings that enable diverse downstream tasks. GrapHist integrates masked autoencoders and heterophilic graph neural networks that are explicitly designed to capture the heterogeneity of tumor microenvironments. We pre-train GrapHist on a large collection of 11 million cell graphs derived from breast tissues and evaluate its transferability across in- and out-of-domain benchmarks. Our results show that GrapHist achieves competitive performance compared to its vision-based counterparts in slide-, region-, and cell-level tasks, while requiring four times fewer parameters. It also drastically outperforms fully-supervised graph models on cancer subtyping tasks. Finally, we also release five graph-based digital pathology datasets used in our study at https://huggingface.co/ogutsevda/datasets , establishing the first large-scale graph benchmark in this field. Our code is available at https://github.com/ogutsevda/graphist .
title GrapHist: Graph Self-Supervised Learning for Histopathology
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2603.00143