Modified fuzzy multi-thresholding algorithm for segmentation of MRI

To overcome the drawbacks of fuzzy multi-level entropic thresholding algorithm, a modified fuzzy multi-level thresholding algorithm for segmentation of MRI is presented in this paper. The algorithm is different from the entropic thresholding algorithm including two aspects. First, maximum variance criteria is used in the proposed algorithm to overcome the limitation of entropic algorithm, which usually requires the resemble distribution of object and background. Second, genetic algorithm is employed to search for the optimal parameters in ours instead of simulated annealing in entropic thresholding algorithm. Besides, the minimum cross-entropy algorithm is applied for masking before segmentation by taking into account the characteristic of MRI. Experimental results show that the proposed method can accurately and efficiently segment the MRI and outperforms the existing ones using the fuzzy entropic thresholding principle.