Content-based Image Retrieval for Computer-aided Detection of Mammographic Masses
Xiangyang Xu · Journal of Chinese Computer Systems · 2008
A content-based image retrieval (CBIR) method is proposed for computer-aided detection (CAD) of mammographic masses. An image database is established.It contains images depicting regions of interest (ROIs) which are known being masses or non-masses but highly suspicious.24 features are extracted from each ROI.By querying the image database,a decision index can be calculated for a ROI to describe in quantity the likelihood that the ROI indeed contains a mass.False positive rates are evaluated under various sensitivity conditions.The experiment results show that the method can achieve an ROC area index AZ=0.841,which is better than the result obtained by the artificial neural network method developed by the Pittsburg University,USA.