PathAI Publications

Characterisation of tumor microenvironment heterogeneity by human-interpretable histopathology-only artificial intelligence in real world diagnostic samples from solid tumors

Written by Admin | Oct 23, 2026, 1:00:00 PM

Introduction

- Tissue and cellular functional states and therapy-related alterations in progressing tumours manifest as structural and spatial changes in individual tumour cells and the cellular and non-cellular components of the tumour microenvironment (TME).

- Haematoxylin and eosin (H&E) stained diagnostic tissue samples reflect such structural manifestations thanks to the unique combination of dyes that label nucleic acids (haematoxylin) and proteins (eosin). This allows detailed visualisation of cancer tissue and cellular and subcellular alterations, as well as visualisation of the spatial distribution of cellular and non-cellular components of the tumour microenvironment (TME).

- However, the precise visual quantification of such alterations is beyond the capacity of human pathologists and has not yet been fully captured by standard artificial intelligence (AI)-based approaches.

- Beyond cellular components, evaluating and quantifying stromal remodelling in the solid tumour microenvironment (TME) has been particularly challenging.

- The distribution and density of stromal fibres provide critical architectural information and can enable or impede invasive tumour growth, immune cell migration, and the effective delivery of systemic and targeted therapies.

Conference

ESMO

View Poster