A Novel Method for Segmentation of CT Head Images

A novel method for CT image segmentation of brain is proposed in this paper, with which, several regions that are suspicious of having pathological changes can be segmented rapidly and effectively. To avoid the affection of extra-cranial highlight pixels in the later processing, row scanning approach is introduced firstly to search for the boundary of the skull. Secondly, combined with morphological operator, intra-cranial area is extracted accurately. Finally, 2D Otsu threshold segmentation is implemented to identify areas of interest and the threshold is obtained through PSO algorithm. The result of simulation demonstrates the efficiency of our method.