Detection of lesions in human brain is an important task aimed at saving precious lives. We propose a segmentation algorithm that is used to segment the lesions of human brain. The novelty of the method of segmentation is the use of random walk algorithm together with entropy maximization. Existing segmentation algorithms using random walk requires user intervention to identify the seed points. We employ entropy maximization to automatically identify the seed points that are to be used in random walk algorithm. The result is a method of segmentation with minimal user intervention. There is also a considerable improvement in segmentation accuracy.
Volume 11 | 04-Special Issue
Pages: 830-835