Self-Supervised AI Discovery of Histomorphological Phenotypes from Routine Mesothelioma Biopsies
- 1School of Cancer Sciences, University of Glasgow, Glasgow, UK; Cancer Research UK Scotland Institute, Glasgow, UK; Queen Elizabeth University Hospital, NHS Grea
Abstract
Accurate subtype diagnosis is essential for guiding therapy and predicting patient outcome in malignant mesothelioma. Most computational pathology models are trained on large tissue images from resection specimens, which maximises information for training but limits model relevance in real-world diagnostic settings where small biopsies are the most usual tissue source. In this work, we assembled a large multicentre cohort of HES- and HPS-stained mesothelioma biopsy slides. We used a self-supervised learning model to evaluate the associations of biopsy-driven morphology patterns with histological subtype, molecular markers, and survival. The discovered histomorphology patterns captured a continuum of tissue phenotypes spanning epithelioid, sarcomatoid, and non-tumour morphologies. Also, patient-level HPC representations achieved excellent performance for distinguishing epithelioid from non-epithelioid mesothelioma (AUC = 0.94) and demonstrated predictive value for immunohistochemistry (IHC) markers. Additionally, HPC-derived features alone achieved performance comparable to established clinical and molecular variables (C-index = 0.65), while integration of HPCs with clinical and marker information improved performance to a C-index of 0.69. Several HPCs were significantly associated with favourable or adverse prognosis and reflected known subtype-specific biological patterns. In conclusion, self-supervised learning can discover interpretable histomorphological phenotypes directly from routine mesothelioma biopsies without further training. These AI-derived phenotypes capture clinically and biologically relevant information, linking tissue architecture to molecular characteristics, histological subtypes, and patient outcomes. The proposed framework provides a thorough evaluation of real-world biopsy data using a pre-trained model, without the need for computationally intensive retraining, and addresses the question of whether SSL-based AI can be deployed out of the box in clinical settings.
Related
- Field-of-view confounding shapes genetic discovery from self-supervised cardiac-imaging phenotypes
- Integrative Bioinformatics and Machine Learning Approaches for Biomarker and Pathway Discovery in Malignant Pleural Mesothelioma
- Genetic discovery using deep learning-derived optic nerve integrity phenotypes
- Hybrid grading systems improve progression risk stratification in non‐invasive papillary urothelial carcinoma and support a histomorphological continuum of disease progression
- Histomorphological changes in low-birth-weight term placentae and association with maternal factors: a comparative cross-sectional study
BibTeX
@misc{seyedshahi2026selfsupervised,
title = {{Self-Supervised AI Discovery of Histomorphological Phenotypes from Routine Mesothelioma Biopsies}},
author = {F. A. Seyedshahi and F. Damiola and R. Sequeiros and F. Forest and A. Scherpereel and K. Yuan and S. Lantuejoul and J. Le Quesne},
howpublished = {bioRxiv},
year = {2026},
month = aug,
doi = {10.64898/2026.08.09.743741},
url = {https://www.biorxiv.org/content/10.64898/2026.08.09.743741v2},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Self-Supervised AI Discovery of Histomorphological Phenotypes from Routine Mesothelioma Biopsies
AU - F. A. Seyedshahi
AU - F. Damiola
AU - R. Sequeiros
AU - F. Forest
AU - A. Scherpereel
AU - K. Yuan
AU - S. Lantuejoul
AU - J. Le Quesne
PY - 2026
DA - 2026/08/09
PB - bioRxiv
DO - 10.64898/2026.08.09.743741
UR - https://www.biorxiv.org/content/10.64898/2026.08.09.743741v2
AB - Accurate subtype diagnosis is essential for guiding therapy and predicting patient outcome in malignant mesothelioma. Most computational pathology models are trained on large tissue images from resection specimens, which maximises information for training but limits model relevance in real-world diagnostic settings where small biopsies are the most usual tissue source. In this work, we assembled a large multicentre cohort of HES- and HPS-stained mesothelioma biopsy slides. We used a self-supervised learning model to evaluate the associations of biopsy-driven morphology patterns with histological subtype, molecular markers, and survival. The discovered histomorphology patterns captured a continuum of tissue phenotypes spanning epithelioid, sarcomatoid, and non-tumour morphologies. Also, patient-level HPC representations achieved excellent performance for distinguishing epithelioid from non-epithelioid mesothelioma (AUC = 0.94) and demonstrated predictive value for immunohistochemistry (IHC) markers. Additionally, HPC-derived features alone achieved performance comparable to established clinical and molecular variables (C-index = 0.65), while integration of HPCs with clinical and marker information improved performance to a C-index of 0.69. Several HPCs were significantly associated with favourable or adverse prognosis and reflected known subtype-specific biological patterns. In conclusion, self-supervised learning can discover interpretable histomorphological phenotypes directly from routine mesothelioma biopsies without further training. These AI-derived phenotypes capture clinically and biologically relevant information, linking tissue architecture to molecular characteristics, histological subtypes, and patient outcomes. The proposed framework provides a thorough evaluation of real-world biopsy data using a pre-trained model, without the need for computationally intensive retraining, and addresses the question of whether SSL-based AI can be deployed out of the box in clinical settings.
ER -