基于神经网络──粒子群优化算法的医学图像分割新方法

A new particle swarm optimization algorithm for medical image segmentation based on neural network

提出了一种基于神经网络──粒子群优化算法的医学图像分割新方法,将粒子群优化算法用于神经网络的学习训练,克服了神经网络收敛速度慢,以及容易收敛到局部最优解的缺点。并与遗传算法进行了比较,结果表明,该方法不仅能更快地收敛于最优解,而且在分割准确率、误差率以及运行时间上也都有了较明显的提高。

A new particle swarm optimization （PSO） algorithm for medical image segmentation based on neural network is put forward. It overcomes both slow convergence speed and local optimization of the neural network. Being compared with the genetic algorithm, it can speedily converge to the optimal solution with improved segment accurate rate, error rate and running time.