具有带通选择性的ICA算法可以改善对于带通时间序列的分离以及对于周期性脑功能响应信号的提取. 因此本文提出的方案可将被估计信号, 如:周期性响应信号以及具有平滑空间分布的脑功能激活区, 的先验特性以特征选择的方式加入ICA算法用以提高对此类信号的估计-with selective ICA algorithm can be improved for the band pass time series, as well as for the separation of brain function cyclical response signal extraction. Therefore this paper, the program can be estimated signals, such as : the cyclical response signal as well as smoothing the spatial distribution of activation of brain function, a priori properties Feature selection to join the ICA algorithm used to enhance the signal of such estimates