BUS-CoT is a multimodal breast-ultrasound dataset designed to cover the full range of histopathology categories while pairing visual evidence with structured diagnostic reasoning. It supports research on clinically grounded chain-of-thought, robust multimodal understanding, and evaluation across diverse lesion types, with public data and reproducible resources released alongside the paper.
A reasoning-rich breast ultrasound resource that connects images, pathology coverage, and clinically grounded chains of thought.