Comparison and evaluation of three optimization algorithms in MRF model for brain tumour segmentation in MRIs
Sahar Yousefi, Reza Azmi, Morteza Zahedi · Iranian Conference on Electrical Engineering · 2011
Summary from only given. MRI brain segmentation plays an increasingly important role in diagnosis and treatment of diseases. Since MRI segmentation manually consumes valuable human resources, a great deal of efforts has been made to automate this process. MRF has been one of the most active research areas of MRI brain segmentation which seeks an optimal label field in a large space. The classical optimization algorithm is Simulated Annealing (SA) that could get the global optimal solution with heavy computation burden. Hence many efforts have been made to obtain the optimal solution in a reasonable time. In this paper, a comparison evaluation of two proposed optimal researching algorithms with the classical MRF for brain tumour segmentation is presented. The first applies a combination of improve genetic algorithm (IGA) and SA, the second uses a hybrid of ant colony optimization (ACO) and gossiping algorithm. The obtained results can assist users to select the appropriate approach for tumour segmentation.