论文部分内容阅读
脑内神经信号的动态传递过程在临床脑疾病诊断和认知科学中得到了越来越多的重视和研究。为求解动态脑磁逆问题,本文提出通过时域子空间求解动态脑磁L2范数解的方法并验证其可行性。具体地,对脑磁测量信号进行奇异值分解,通过右奇异向量矩阵构造源信号的时域子空间,将测量信号和待求源信号投影到时域空间进行求逆,然后将解反投影到解空间重建出动态脑磁源信号。与施加时域约束的双正则化方法相比,本文方法计算量大大降低,不同噪声水平下本文方法的均方误差都小于双正则化方法,而且重建出的源信号与仿真信号吻合更佳,信噪比更高。
The dynamic transmission of neural signals in the brain has gained more and more attention and research in clinical brain disease diagnosis and cognitive science. In order to solve the dynamic magnetoencephaltonic inverse problem, this paper proposes a method to solve the dynamic magnetoencephaloreceptor L2 norm by time-domain subspace and verifies its feasibility. Specifically, the brain magnetic measurement signals are singularly-valued decomposed, the time-domain subspace of the source signal is constructed by a right singular vector matrix, the measurement signal and the signal to be sought are projected into the time-domain space for inversion, and then the inverse projection is performed on Solve the space to reconstruct the dynamic brain magnetic signal. Compared with the bi-regularization method with time-domain constraint, the computational complexity of this method is greatly reduced. The mean square error of the proposed method is less than that of the bi-regularization method under different noise levels, and the reconstructed signal matches the simulation signal better. Higher signal to noise ratio.