Identification of Source Signals by Estimating Directional Index of Phase Coupling in Multivariate Neural Systems

Haojie Xu, Haijun Shan, Bin He, Shanan Zhu

Research output: Contribution to journalArticlepeer-review


In the field of biomedical engineering, identification of source signals between multichannel signals is an important technique for measuring the functional connectivity of a variety of brain states and pathological cases such as seizures. It is likely to occur in a weak coupling condition, where the neural actions interact with each other on the level of phase, whereas their amplitudes remain practically uncoupled. Hence, source signal identification by analyzing the directional interactions from a phase-coupled complex neural network is of interest. This study proposes a technique for identifying active source signals by estimating the directional effects of phase coupling in multivariate neural systems. In the proposed method, the directional index of the phase coupling is developed from the framework of phase noise situation and the Kuramoto model. The proposed method is evaluated and compared with the well-known adaptive directed transfer function (ADTF) method via simulations of various chaotic oscillatory processes and interictal spikes data obtained from an epileptic patient. The results indicate that the proposed source identification technique outperforms the conventional ADTF method under the weak coupling condition and that it can discriminate between direct and indirect connectivity.

Original languageEnglish (US)
Pages (from-to)273-281
Number of pages9
JournalJournal of Medical and Biological Engineering
Issue number2
StatePublished - Apr 1 2016

Bibliographical note

Publisher Copyright:
© 2016, Taiwanese Society of Biomedical Engineering.


  • Directional connectivity
  • Epilepsy
  • Interictal spikes
  • Phase coupling
  • Source signal identification


Dive into the research topics of 'Identification of Source Signals by Estimating Directional Index of Phase Coupling in Multivariate Neural Systems'. Together they form a unique fingerprint.

Cite this