Quantum and classical genetic algorithms for multilevel segmentation of medical images: A comparative study

Abstract

In this paper, we propose a multilevel segmentation methods of medical images based on the classical and quantum genetic algorithms. The Genetic Algorithm (GA) uses a binary coding while the Quantum Genetic Algorithm (QGA) uses the qubit encoding of individuals. The two evolutionary algorithms are employed to maximize efficiently Rényi, Masi and Shannon entropies for the purpose of multi-objects segmentation of medical images. The Particle Swarm Optimization algorithm (PSO) was also used for comparison reasons. The segmentation quality of the nine proposed approaches is assessed by means of the prevailing indices PSNR, SSIM and FSIM. The numerical results and the comparative study were carried out on a sample of twenty medical images. It was shown that the QGA outpaces the GA, and the PSO outperforms significantly the both algorithms in the optimization task. Finally, it was found that the Rényi entropy is more suitable for the purpose of medical image multilevel thresholding.