New AI model enhances detection of colorectal liver metastases

An innovative AI-based recognition model for detecting colorectal liver metastases during contrast-enhanced intraoperative ultrasonography was developed. By integrating the basic recognition model and a subtraction model, a combination model demonstrated superior accuracy, reaching 96.5%. Overall, the combination model achieved an exceptional area under the curve (AUC) of 0.99, significantly surpassing individual models, which achieved 89.4% and 86.6% accuracy. This advancement could streamline tumor detection, improving surgical outcomes.