電子情報通信学会総合大会講演要旨
A-2-3
Liver lesion retrieval based on multiphase medical volumes using sparse coding based codebook model
◎Jian WANG・Xian-hua Han(Ritsumeikan Univ.)・Yingying Xu・Lanfen Lin・Hongjie Hu・Chongwu Jin(Zhejiang Univ.)・Yen-Wei Chen(Ritsumeikan Univ.)
The content based medical image retrieval plays an important role in the medical diagnosis and experience sharing, which can assist the resident doctors or even medical experts making better decisions. To retrieve the similar cases to the query sample, the extraction of inherent feature representation for medical images is needed. This study explores a codebook-based feature representation. A novel codebook-based model using sparse coding is employed and the raw patches are extracted as local descriptors. The proposed method is applied to retrieval of the liver lesions using multiphase medical volumes. Experiments validate that the promising retrieval performance can be achieved