Mapping the histomorphological landscape across mouse tumours with self-supervised AI models

Dr Ke Yuan & Prof John Le Quesne

Labs: AI for Cancer Research & Deep Phenotyping of Solid Tumours
Duration: 4 years, starting October 2025
Closing Date: Friday 14 March 2025
Interviews for this position will take place in April 2025

Background

Histological assessment of mouse tumor phenotypes plays a crucial role in functional studies. Traditionally, this analysis has relied on manual examination, which is insufficient for modern experimental scales that require the evaluation and quantification of hundreds or even thousands of histology slides. Self-supervised AI models in pathology present an opportunity to significantly improve the efficiency and accuracy of phenotypic assessments in mouse models.

Our group has recently developed the Histomorphological Phenotype Learning (HPL) framework, a self-supervised AI tool designed to detect recurring patterns in pathology slides. This approach has already been successfully applied to several human cancers, uncovering a landscape of previously underappreciated tumor phenotypes. Building on this success, we aim to apply self-supervised AI models to systematically map histologic patterns across multiple cancer types in mouse models. These AI-driven models will not only enhance our understanding of tumor phenotypes but will also facilitate rapid and scalable quantification of histological data, applicable to future experiments as soon as the images are generated.

Research Question

The primary goal of this project is to develop self-supervised AI foundation models using mouse pathology slides. These models will be trained to predict molecular characteristics, including mutation profiles, transcriptomic signatures, and proteomic data. Furthermore, we will integrate these multimodal data with human datasets for the purpose of target validation and understanding disease mechanisms. Through this work, we seek to advance the field of pathology by enabling more precise and scalable histological analysis in both preclinical and clinical settings.

Skills/Techniques that will be gained

1. Programming skills in training, validation and testing state-of-the-art deep learning models.
2. State-of-the-art methods in computer vision and medical image analysis
3. Biological interpretation and histological assessment of H&E and mIF images
4. Broad understanding of the translational value of cutting-edge data and technology in cancer sciences.

For questions regarding the application process, PhD programme/studentships at the CRUK Scotland Institute or any other queries, please contact phdstudentships@beatson.gla.ac.uk.

Closing date: Friday 14 March 2025

Applications are open to all individuals irrespective of nationality or country of residence.

 APPLY HERE

Relevant Publications

Claudio Quiros A et al. Mapping the landscape of histomorphological cancer phenotypes using self-supervised learning on unannotated pathology slides. Nature Communications. 2024 Jun 11;15(1):4596. doi: 10.1038/s41467-024-48666-7. PMID: 38862472.

Chen R et al. Towards a general-purpose foundation model for computational pathology. Nature Medicine 2024 Mar;30(3):850-862. doi: 10.1038/s41591-024-02857-3

Pan X, et al The artificial intelligence-based model ANORAK improves histopathological grading of lung adenocarcinoma. Nature Cancer. 2024 Feb;5(2):347-363. doi: 10.1038/s43018-023-00694-w. Epub 2024 Jan 10. PMID: 38200244

AbdulJabbar K et al  Bridging clinic and wildlife care with AI-powered pan-species computational pathology. Nature Communications. 2023 Apr 26;14(1):2408. doi: 10.1038/s41467-023-37879-x.