基于线性约束最小方差的脑磁源定位特性研究

Investigation on MEG Source Localization with Linear Constrained Minimum Variance

波束形成是一种广泛运用于脑磁信号的偶极子溯源方法,其定位结果的准确度是目前研究的一个关键点。基于电流偶极子模型,以相关系数和定位误差作为评价标准,研究了不同头模型、不同伪迹噪声对线性约束最小方差定位算法的影响。通过计算机软件,在脑内已知位置设定已知源信号,采用不同头模型进行前向问题计算并叠加不同噪声,对模拟的真实脑磁信号进行逆问题的求解,进行源定位与源信号重构。仿真结果表明,在叠加相同噪声的情况下,采用不同头模型在较低信噪比下对算法的影响有一定差异,而在信噪比高于-10分贝的条件下,则对算法几乎没有影响,能达到较好的定位效果。在采用相同头模型的情况下,叠加不同类型的噪声伪迹所产生的影响各不相同,其中高斯白噪声产生的影响最大,有色噪声次之,基线漂移产生的影响最小。

Beamforming is widely used in the dipole sourcing for magnetoencephalography （MEG） signals, and the accuracy of localiza- tion result is the key of the current research. Based on the current dipole source model, the influence of different head models and artifacts on the Linear Constrained Minimum Variance （LCMV） sourcing has been discussed, taking the correlation and the localization bias as the valuation criteria. The source signal has been simulated using computer software and the forward problem has been solved based on different head model,then the inverse problem is calculated to localize the sources and reconstruct the signals under different noise envi- ronments. The simulation suggests that under the same noise environment, the localization result has been affected by the head model while the signal to noise ratio is low,however the accuracy of localization result is more reliable and not related to the head model when the SNR is higher than -10dB. In the case of the same head model,the types of artifacts lead to different results,where Gaussian white noise has the greatest influence, colored noise is second, and baseline wander artifact has the least influence.