Improvement of brain lesions detection using information fusion approach

Automatic segmentation of brain lesions, such as multiple sclerosis in MRI images, is a complex operation. One of the main difficulties is to optimize the dilemma between the false positives and false negatives present in the segmented image. We propose here a new approach to this problem. The idea is to exploit the complementary results from different segmentation algorithms as well as a priori knowledge to reduce false positives. The method starts with modeling inaccuracy about the borders of the segmented regions. The logic rules are then defined in order to combine the white matter image and lesions within the framework of evidence theory. The results show that brain lesion detection is substantially improved using this data fusion approach.