Artificial intelligence-powered copilots for precision diagnosis and surgical assessment of histological growth patterns in resectable colorectal liver metastases: a prospective study

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 developed a Transformer-based deep learning model, COFFEE, for the precise classification of colorectal cancer subtypes using whole-slide images (WSIs) from 431 patients diagnosed with colorectal cancer liver metastasis. The model was pretrained using DINO on 1442 WSIs from the TCGA-COAD cohort, utilizing a Vision Transformer (ViT) architecture to extract 384-dimensional feature vectors from 256 × 256 pixel patches. The proposed model integrates a Transformer-based Multiple Instance Learning (TransMIL) framework, which effectively aggregates spatial and morphological information through multi-head self-attention and Pyramid Position Encoding Generator (PPEG) modules. This design enables efficient handling of large instance sequences within WSIs, allowing for accurate binary and four-class classification. The model was validated on 972 WSIs from a recent dataset, demonstrating its robustness and clinical applicability. After testing the model with internal and prospective cohorts, a direct comparison between conventional and AI-assisted pathology assessment was also performed.

A total of 431 patients were included in three cohorts: training ( n = 297), testing ( n = 104), and prospective ( n = 30). Desmoplastic tumors were associated with longer overall survival (OS, 53.6 vs. 31.9 months, P = 0.002) and progression-free survival (PFS, 25.2 vs. 10.7 months, P < 0.001) compared to non-desmoplastic tumors. The COFFEE binary classification model achieved high predictive performance with area under the ROC curve (AUC) values of 0.961 in the training, 0.935 in the testing, and 1.000 in the prospective cohort. The four-class model also showed strong performance, with AUCs of 0.961 and 0.966 in the training and testing cohorts, and 0.985 in the prospective cohort. AI-assisted models helped junior pathologists achieve an accuracy of 94.7% (vs. 85.9%) and reduced diagnostic time by 36%, improving both accuracy and speed.

This study developed an AI model for HGP classification in colorectal cancer liver metastasis, achieving high accuracy in both binary classification and four-class classification models. The model demonstrated potential for improving diagnostic precision and guiding post-surgery treatment strategies, with AI-assisted pathologists surpassing traditional methods in a prospective cohort.