Robust Multicentre Detection and Classification of Colorectal Liver Metastases on CT: Application of Foundation Models

TL;DR

This study presents a foundation model-based AI pipeline for accurate, robust, and interpretable detection and classification of colorectal liver metastases on multi-centre CT scans, demonstrating high performance and clinical utility.

Contribution

The paper introduces a novel foundation model approach tailored for multi-centre CRLM detection and classification, with integrated uncertainty quantification and explainability.

Findings

Achieved an AUC of 0.90 for classification and 69.1% lesion detection overall.

Improved performance to AUC 0.91 and accuracy 0.86 when excluding uncertain cases.

Detected lesions with increasing accuracy across size quartiles, up to 98% for largest lesions.

Abstract

Colorectal liver metastases (CRLM) are a major cause of cancer-related mortality, and reliable detection on CT remains challenging in multi-centre settings. We developed a foundation model-based AI pipeline for patient-level classification and lesion-level detection of CRLM on contrast-enhanced CT, integrating uncertainty quantification and explainability. CT data from the EuCanImage consortium (n=2437) and an external TCIA cohort (n=197) were used. Among several pretrained models, UMedPT achieved the best performance and was fine-tuned with an MLP head for classification and an FCOS-based head for lesion detection. The classification model achieved an AUC of 0.90 and a sensitivity of 0.82 on the combined test set, with a sensitivity of 0.85 on the external cohort. Excluding the most uncertain 20 percent of cases improved AUC to 0.91 and balanced accuracy to 0.86. Decision curve analysis showed clinical benefit for threshold probabilities between 0.30 and 0.40. The detection model identified 69.1 percent of lesions overall, increasing from 30 percent to 98 percent across lesion size quartiles. Grad-CAM highlighted lesion-corresponding regions in high-confidence cases. These results demonstrate that foundation model-based pipelines can support robust and interpretable CRLM detection and classification across heterogeneous CT data.