本研究发展了一种基于相关系数分解的脑电相干源成像方法。具体给出了多道脑电的相关系数分解算法和相应的相干源成像方法。以能量成像方法为对照,证实了方法对相干但能量水平不同的源组合的敏感性以及抗噪能力。结果表明,这种方法成像敏感度高。能对多个相干源成像,且具有较强的抗噪声性,因而有明显的理论意义和应用价值。
A novel algorithm, termed multivariate correlation coefficient decomposition, is proposed to map the coherent sources in human brain. In this work we demonstrate how to apply correlation coefficient decomposition algorithm to multi-channel scalp EEG and illustrate the corresponding methods for mapping coherent sources. Compared with energy mapping, this algorithm was sensitive to the combination of coherent sources with different energy levels. Computer simulation and VEP data tests show that the presented algorithm has high sensitivity in mapping coherent sources, it accurately map multiple coherent sources and was robust to noise.