GA-BW based HMM in Brain Image Segmentation

Abstract

Image segmentation is an important preprocessing step in a sophisticated and complex image processing algorithm. In segmenting real-world images, the cost of misclassification could depend on the true class. For example, in a two-class (negative or positive class) problem, the cost of misclassifying positive to negative class could not be equal to that of misclassifying negative to positive class. However, existing algorithms do not take into account the unequal misclassification cost. Here, we introduce a procedure to minimize the misclassification cost with class-dependent cost. The procedure assumes the hidden Markov model (HMM) which has been popularly used for image segmentation in recent years. In this proposed method   Baum-Welch (B-W) Algorithm is used to calculate the HMM model parameters. However, the  B-W  algorithm  uses  an  initial  random  guess  of  the  parameters,  therefore, after  convergence the output tends  to be close to this  initial value of the  algorithm,  which  is  not necessarily  the  global  optimum  of the  model  parameters.  To achieve an optimum result   Genetic  Algorithm (GA)  combined with Baum-Welch  (GA-BW)  is  proposed   and the  idea  is to  use  GA  exploration  ability  to  obtain  the  optimal  parameters  within  the solution space. By using this proposed method, brain tumor region and non tumor region is segmented and classified within the state of art.