Monarch butterfly optimization algorithm for computed tomography image segmentation

Abstract

In the medical field, image segmentation provides important information for surgical planning and registration, and thus demands accurate segmentation. In order to improve the effectiveness and the threshold accuracy of segmentation, researchers have tended to use a metaheuristic algorithm as the operational algorithm to achieve better exploitation and exploration of the search space and to solve many different medical image problems in an effective manner. This research applies the monarch butterfly optimization (MBO) algorithm for image segmentation at multiple threshold values. To evaluate the performance of the implemented MBO algorithm, a comparison is made with the brute-force (i.e. Otsu) algorithm and two metaheuristic algorithms (i.e. Darwinian particle swarm optimization (DPSO) and fractional-order DPSO). In addition, the quality structural similarity index matrix and the peak signal-to-noise ratio are used to evaluate the accuracy of the resultant segmented images. The experimental results show the advantage of using the MBO algorithm for medical image segmentation in terms of accuracy and speed. Regards the accuracy, MBO algorithm produced an exact match at thresholds 1 and 2 and a very close match at thresholds 3 and 4. Regards the speed, the average of the execution time for threshold 1 was 0.24187 s, for threshold 2 was 0.33831 s, for threshold 3 was 0.95967 s and for threshold 4 it was 1.15308 s.