Computer Aided Detection of Cavernous Malformation in T2-Weighted Brain MR Images

Abstract

Cavernous malformation or cavernomas is abnormal development of brain blood vessels and affects a large population (0.5%) all over the world. It could cause seizures, vision or memory problems as the malformations may happen at different locations of brain. Radiologists usually analyze brain magnetic resonance (MR) images to detect cavernomas. However, automatic detection of cavernomas by computer has not been investigated enough. This paper proposes a computer aided cavernomas detection method based on MR images analysis. The proposed method includes three steps: brain extraction based on deformable contour (to remove the nonbrain tissues from image), template matching (to find suspected cavernomas regions) and post-processing (to get rid of false positives based on size, shape and brightness information). The performance of the proposed technique is evaluated and a sensitivity of 0.82 is obtained after testing.