如何有效地从头表记录电位中准确定位脑电源的真实活动位置是神经认知脑功能研究中的一个关键问题.本文在FOCUSS算法迭代基础上,从脑神经活动的局部稀疏性出发,提出了一种新的脑功能成像方法,在该算法中,通过把稀疏性的lp模约束加入到修改的FOCUSS算法的迭代过程中,使算法可以有效地收敛于真实的稀疏源活动位置.利用该方法对随机系统、三层球模型及真实头模型确定的稀疏欠定系统进行了求解模拟实验,结果显示了该方法在求解欠定系统及EEG源定位时具有良好的稳健性.
How to localize the neural active source areas effectively and precisely from the scalp EEG recordings is a critical issue for clinical neurology and cognitive neuroscience. In this paper, based on the sparsity assumption of brain activities, proposed is a novel iterative EEG source imaging algorithm, which is a modified FOCUSS iteration procedure combined with lp norm sparse constraints. In this algorithm, the sparse constraint of solution, lp norm, is integrated into the modified iterative procedure of FOCUSS, and it is the renewed weight subjected to the lp norm constraints that forces the solution to converge to the active sparse source position effectively. The conducted simulation studies for a random underdetermined system, a 3-shell sphere model and a realistic head model all showed its validation for solving of underdetermined system and EEG source imaging.