Researchers have introduced CAPTOR, a vision-language foundation model designed for oncologic CT interpretation. The model targets scans of the chest, abdomen, and pelvis, a common imaging protocol for cancer staging and follow-up. According to the arXiv preprint, the work aims to develop and evaluate this deep-learning system.

The abstract describes CAPTOR as a deep-learning vision-language foundation model, meaning it is trained to process both imaging and textual data. The study's stated purpose is to develop and evaluate the model for oncologic CAP CT interpretation. As the abstract is an early announcement, no results or performance metrics are included in the available text.