Preliminary study on the automated detection of tumors using various unenhanced breast MR images: false positive reduction with signal intensities

Abstract

Breast cancer incidence tends to rise. Dynamic MRI examination using contrast agent is conducted for the diagnosis of breast cancer. However, contrast agent may harm the screening examinees. Here, MRI provides the anatomical and functional information of tumors without contrast agent by using various sequences. In this study, we developed an automated method for the detection of tumors using unenhanced MR images. The breast region is extracted from T1WI and the tumor candidates are detected from DWI. The false positives among the initial candidates are eliminated by the rule-based classifier. We applied the proposed method to 54 cases; the detection sensitivity was higher than 70%.