Methods Inf Med 2018; 57(03): 122-128
DOI: 10.3414/ME17-01-0151
Focus Theme – Original Article
Schattauer GmbH

A Robust Method with High Time Resolution for Estimating the Cortico-Thalamo-Cortical Loop Strength and the Delay when Using a Scalp Electroencephalography Applied to the Wake-Sleep Transition

Ikuhiro Yamaguchi
1  Graduate School of Education, The University of Tokyo, Tokyo, Japan
,
Akifumi Kishi
1  Graduate School of Education, The University of Tokyo, Tokyo, Japan
,
Fumiharu Togo
1  Graduate School of Education, The University of Tokyo, Tokyo, Japan
,
Toru Nakamura
2  Graduate School of Engineering Science, Osaka University, Osaka, Japan
,
Yoshiharu Yamamoto
1  Graduate School of Education, The University of Tokyo, Tokyo, Japan
› Author Affiliations
This work was supported by JSPS KAKENHI Grant Number 15K01499.
The authors thank Professor Benjamin H. Natelson for providing his data for this analysis and for his useful discussions.
Further Information

Publication History

received: 18 December 2017

accepted: 25 January 2018

Publication Date:
02 May 2018 (online)

Summary

Objectives: This study aimed to describe a robust method with high time resolution for estimating the cortico-thalamo-cortical (CTC) loop strength and the delay when using a scalp electroencephalography (EEG) and to illustrate its applicability for analyzing the wake-sleep transition.

Methods: The basic framework for the proposed method is the parallel use of a physiological model and a parametric phenomenological model: a neural field theory (NFT) of the corticothalamic system and an autoregressive (AR) model. The AR model is a “stochastic” model that shortens the time taken to extract spectral features and is also a “linear” model that is free from the local-minimum problem. From the relationship between the transfer function of the AR model and the transfer function of the NFT in the low frequency limit, we successfully derived a direct expression of CTC loop strength and the loop delay using AR coefficients.

Results: Using this method to analyze sleep-EEG data, we were able to clearly track the wake-to-sleep transition, as the estimated CTC loop strength (c 2) decreased to almost zero. We also found that the c 2-distribution during nocturnal sleep is clearly bimodal in nature, which can be well approximated by the superposition of two Gaussian distributions that correspond to sleep and wake states, respectively. The estimated loop delay distributed ∼0.08 s, which agrees well with the previously reported value estimated by other methods, confirming the validity of our method.

Conclusions: A robust method with high time resolution was developed for estimating the cortico-thalamo-cortical loop strength and the delay when using a scalp electroencephalography. This method can contribute not only to detecting the wake-sleep transition, but also to further understanding of the transition, where the cortico-thalamo-cortical loop is thought to play an important role.