为了保持原始功能磁共振成像(function Magnetic Resonance Ima-ging,fMRI)数据的空间结构实现噪声体素移除,提高聚类的效果,提出了一种基于遗传算法的简易优化算法.以广义线性模型(Generalize Linear Model,GLM)方法获取的数据作为大脑视觉刺激的真实激活模板,基于遗传算法对5位被测者的fMRI数据图像进行分析.取不同阈值(0.1～0.9)时交叉率和变异率为0且具有唯一的最优值为模糊C均值聚类算法(Fuzzy C-Means,FCM)结果,依据真实模板验证聚类结果的准确度.结果表明,相比原始的FCM方法,改变阈值的大小可以使FCM聚类结果的准确度得到有效提高.通过简易优化遗传算法,可以确定最佳阈值为0.6.
Research on Simple Optimization and Threshold Determination Based on Genetic Algorithm
In order to maintain the spatial structure of the original fMRI data,remove noisy voxels,and improve clustering performance,a simple optimization algorithm based on genetic algorithm is proposed.Using the data obtained by the GLM method as the real activation template for brain visual stimuli,the fMRI data of five subjects are analyzed based on genetic algorithm.The fuzzy c-means(FCM)clustering results with different thresholds(0.1-0.9),where the crossover rate and mutation rate are 0 and have a unique optimal value,are used to verify the accuracy of the clustering results based on the real template.The results indicate that compared to the original FCM method,changing the threshold size can effectively improve the accuracy of FCM clustering results.By using a simple optimized genetic algorithm,the optimal threshold can be determined to be 0.6.
NETL
NSTL
万方数据
