基于混沌优化算法的MUSIC脑磁图源定位方法

如何利用实验测得的脑磁图数据准确定位脑磁图源的真实活动位置是脑功能研究和临床应用中的一个关键问题.在脑磁活动源定位问题中,多信号分类算法是被广泛研究和采用的一类方法.为了克服多信号分类算法及其改进算法--递归多信号分类算法全局扫描时速度太慢的缺点,提出了一种基于混沌优化算法的脑磁图源定位新方法.该方法利用混沌运动遍历性的特点估计目标函数的全局最大值,进行初步的脑磁图源定位;然后,在小范围内结合网格的方法,进一步进行精确的定位.实验结果表明,此方法可实现多个脑磁图源的定位,并且定位速度大大加快,同时又能达到所要求的定位精度.

Music Localization for MEG Sources Based on Chaos Optimization Algorithm

How to localize the neural activitation sources effectively and precisely from the magnetoencephalographic recordings is a critical issue for the clinical neurology and the study on brain functions. Multiple signal classification algorithm and its extension which is referred to as recursive multiple signal classification algorithm are widely used to localize multiple dipolar sources from the magnetoencephalographic data. The shortage of these algorithms is that they run very slowly when scanning a three-dimensional head volume globally. In order to solve this problem, a novel magnetoencephalographic source localization method based on chaos optimization algorithm is proposed. This method uses the property of ergodicity of chaos to estimate the rough source locations as the arguments which are close to the global maximum of the cost function, then, combining with grids in small areas, the accurate dipolar source localization is performed. Experimental results show that this method can localize multiple dipolar sources easily. The speed of source localization can be improved greatly and the accuracy is satisfactory.