Article,

Brain Tissues Segmentation in MR Images based on Level Set Parameters Improvement

, , and .
Applied Mathematics and Sciences: An International Journal (MathSJ), 1 (3): 53-63 (December 2014)

Abstract

this paper presents a new image processing technique for brain tissue segmentation, precisely,in order to recognize brain diseases. Automatic level set(ALS) is a powerful method for segmenting brain tissues in MR images that uses spatial Fuzzy C-Means (SFCM) to set initial contour near the object’s boundaries in order to increasing the speed of algorithm. Themethod efficiency depends on selecting the optimized amounts of controlling parameter. In this paper, the ALSis improved by optimal regulating of controlling parameters. The proposedmethod contains two phases. In the first phase,the initial contour of ALS determined via the SFCM and image features are extracted. Then, the optimal controlling parameters of ALS are determined by a genetic algorithm.By applying image features and optimal controlling parameters to the generalized regression neural network(GRNN), a neural system is trained. In the second phase, the initial contour is specified and image features are extracted as inputs to trained neural network from phase1. Thus, the outputs of neural network are used as ALS controlled parameters. The results show that the accuracy of proposed ALS is improved about 1.4 % with respect to the ALS method. The proposed ALS not only retains the speed but also has a higher accuracy.

Tags

Users

  • @journalmathsj

Comments and Reviews