BUSGen introduces the first foundational generative model for breast ultrasound image analysis, addressing data scarcity and privacy concerns by generating realistic, diverse synthetic data. This model improves performance across various breast cancer screening, diagnosis, and prognosis tasks, often surpassing human expert capabilities.
This research from Peking University and Yizhun Medical AI Co., Ltd. introduces TAILOR, a pipeline that leverages knowledge-driven AI-generated data to enhance breast ultrasound diagnoses. The approach effectively addresses data scarcity for rare conditions and provides an interpretable diagnostic model, demonstrating superior accuracy compared to conventional models and average human radiologists, especially for challenging cases like Ductal Carcinoma In Situ.
There are no more papers matching your filters at the moment.