作者: Xiaodan Liang , Qingxing Cao , Rui Huang , Liang Lin
DOI: 10.1109/ISBI.2014.6868087
关键词:
摘要: The aim of this study is to provide an automatic computational framework assist clinicians in diagnosing Focal Liver Lesions (FLLs) Contrast-Enhancement Ultrasound (CEUS). We represent FLLs a CEUS video clip as ensemble Region-of-Interests (ROIs), whose locations are modeled latent variables discriminative model. Different types characterized by both spatial and temporal enhancement patterns the ROIs. model learned iteratively inferring optimal ROI optimizing parameters. To efficiently search ROIs, we propose data-driven inference algorithm combining effective pruning. experiments show that our method achieves promising results on largest dataset literature (to best knowledge), which have made publicly available.