Abstract: Medical Image Segmentation is a hot topic in the community of medical images analysis. The traditional genetic algorithm is sometimes inaccurate and instable when it is used in searching the best solutions of some functions. To solve the problem, an Otsu Dual-threshold Value Method based on parallel genetic algorithm for Medical Image Segmentation is proposed. In the algorithm, evolution is performed among different subgroups in parallel. The avoidance of premature convergence of single-species evolutionary process improves the convergence efficiency of the algorithm. The thresholds searching results for 100 times show that the algorithm presented in this paper can not only find better solutions, but also be more stable and accurate than the traditional genetic algorithm. Its convergence is improved more quickly than that of the single-species genetic algorithm.

Key words: medical image, Otsu, threshold, genetic algorithm