Region based fuzzy clustering for automated brain segmentation
Siamak Ardekani, Hooshang Kangarloo, Usha S. Sinha · 2002
A fully automated, fast and accurate method for segmenting contrast enhanced T1 weighted MR head images into brain and non-brain has been developed. The algorithm employs information acquired from the smoothed MR intensity histogram to define thresholds that can be used to first remove the background noise and second segment head mask into smaller regions. A fuzzy clustering technique was then adopted to classify the regions that were obtained from intensity thresholding and morphological operations into brain and non-brain. The algorithm performs successfully both on normal and abnormal MR brain volumes with high intensity space occupying lesions. The algorithm was verified on 10 axial post-contrast T1 weighted images by computing the similarity index for the manually and automatically outlined brain images. The mean similarity index was 0.944 (/spl plusmn/0.0094 SD). The average elapsed time to perform whole process on a Pentium III processor was 9.77 (7.27-12.22) minutes.