Graef Andreas, Hartmann Manfred, Deistler Manfred, Kluge Tilmann
AIT Austrian Institute of Technology GmbH, Safety and Security Department, Vienna, Austria.
Annu Int Conf IEEE Eng Med Biol Soc. 2009;2009:4743-6. doi: 10.1109/IEMBS.2009.5334186.
In this paper we assess a dependency measure for multivariate time series termed Extrinsic-to-Intrinsic-Power-Ratio (EIPR) using two different signal models. In a comparison with Partial Directed Coherence (PDC) we show that both measures correctly identify imposed couplings, but that limitations of the PDC do not affect EIPR. Moreover, EIPR is successfully used for the localization of the seizure onset zone in ECoG recordings from two epilepsy patients, given the exact seizure onset time. The electrodes identified by the proposed method are in excellent accordance with the clinical findings.
在本文中,我们使用两种不同的信号模型评估了一种针对多元时间序列的相关性度量,称为外在与内在功率比(EIPR)。与部分定向相干性(PDC)相比,我们表明这两种度量都能正确识别施加的耦合,但PDC的局限性不会影响EIPR。此外,在已知癫痫发作确切起始时间的情况下,EIPR成功用于定位两名癫痫患者的脑皮质电图(ECoG)记录中的癫痫发作起始区。所提出的方法识别出的电极与临床发现高度吻合。