Deep Learning-Based Analysis of Colorectal Cancer Pathology Images: An Innovative Approach for Predicting Colorectal Cancer Subtypes

Colorectal cancer (CRC) is a leading cause of mortality in China, with metastasis significantly contributing to poor outcomes. Histopathological growth patterns (HGPs) in colorectal liver metastasis (CRLM) provide vital prognostic insights, yet the limited number of pathologists highlights the need for auxiliary diagnostic tools. Recent advancements in artificial intelligence (AI) have demonstrated potential in enhancing diagnostic precision, prompting the development of specialized AI models like COFFEE to improve the classification and management of HGPs in CRLM patients. This study aims to develop and validate a Transformer-based deep learning model, COFFEE, for the classification of colorectal cancer subtypes using whole slide images (WSIs) from patients diagnosed with colorectal cancer liver metastasis. The model is pre-trained using self-supervised learning (DINO) on WSIs from the TCGA-COAD cohort, utilizing a Vision Transformer (ViT) architecture to extract 384-dimensional feature vectors from 256×256 pixel patches. The COFFEE model integrates a Transformer-based Multiple Instance Learning (TransMIL) framework, incorporating multi-head self-attention and Pyramid Position Encoding Generator (PPEG) modules to aggregate spatial and morphological information. The study includes training, testing, and prospective validation cohorts and evaluates the performance of the model in both binary and multi-class classification settings, as well as its potential to assist pathologists in clinical workflows.