SALSA achieves precise and fully automated identification and delineation of liver cancer on CT images, facilitating more accurate quantification of tumor burden, a critical factor in cancer prognosis and management. The model achieved superior accuracy in tumor identification and volume quantification, outperforming state-of-the-art tools and exceeding the inter-reader agreement among three expert radiologists. SALSA holds promise for improving cancer care, enhancing patient outcomes, and increasing healthcare system efficiency by providing rapid, consistent, and reliable data on liver cancer count and volume, supporting both clinical trial endpoints and decision-making in clinical practice.

Limitations

• The masks considered as ground truth were assessed by only one experienced radiologist per scan, introducing variability as has been proven in the substudy.
• The criteria for the segmentation for the external validation cohorts were undisclosed (lesions smaller than 1 cm in diameter were delineated in the ground truth). All the development and test datasets were segmented following RECIST guidelines. Therefore, lesions smaller than 1 cm in diameter are not delineated by SALSA.