Extraction of brain regions affected by Alzheimer disease via fusion of brain multispectral MR images

Abstract

This paper proposes a new method for extraction of regions affected by Alzheimer's disease from multispectral medical images. In this method, first two modals of magnetic resonance images are fused to achieve an image with high information content. Statistical features of fused image are extracted and then are grouped into three clusters with the help of an unsupervised algorithm to perform initial segmentation. Labeling members of clusters and rearranging image yields final image. Results of quantitative analysis prove combination of fusion and segmentation to result in an image with higher values of quantitative metrics and better visual outcome.