Recurrent Deep Neural Networks for Real-Time Sleep Stage Classification From Single Channel EEG.

Erik Bresch,Gary Garcia-Molina,Ulf Großekathöfer

doi:10.3389/fncom.2018.00085

Abstract

Objective: We investigate the design of deep recurrent neural networks for detecting sleep stages from single channel EEG signals recorded at home by non-expert users. We report the effect of data set size, architecture choices, regularization, and personalization on the classification performance.Methods: We evaluated 58 different architectures and training configurations using three-fold cross validation.Results: A network consisting of convolutional (CONV) layers and long short term memory (LSTM) layers can achieve an agreement with a human annotator of Cohen's Kappa of ~0.73 using a training data set of 19 subjects. Regularization and personalization do not lead to a performance gain.Conclusion: The optimal neural network architecture achieves a performance that is very close to the previously reported human inter-expert agreement of Kappa 0.75.Significance: We give the first detailed account of CONV/LSTM network design process for EEG sleep staging in single channel home based setting.

Highlights

SLEEP can be formally defined as a state of reversible disconnection from the environment characterized by quiescence and reduced responsiveness usually associated with immobility
We investigate the design of deep recurrent neural networks for detecting sleep stages from single channel EEG signals recorded at home by non-expert users
A network consisting of convolutional (CONV) layers and long short term memory (LSTM) layers can achieve an agreement with a human annotator of Cohen’s Kappa of ∼0.73 using a training data set of 19 subjects

Summary

Introduction

SLEEP can be formally defined as a state of reversible disconnection from the environment characterized by quiescence and reduced responsiveness usually associated with immobility. The precise function of sleep remains to be elucidated, it appears that sleep primarily benefits the brain (Cirelli and Tononi, 2008). Sleep is of the brain, by the brain, and for the brain (Hobson, 2005); not surprisingly the brain activity during sleep undergoes striking changes compared to that during wakefulness. Non-rapid eye movement sleep (NREM) and rapid eye movement sleep (REM) cyclically alternate with a periodicity of approximately 90 min. REM and NREM sleep occupy ∼20 and 80% of total sleep time, respectively. NREM sleep includes lighter stages N1 and N2 and deep sleep N3 ( known as slow wave sleep). The sleep process can be characterized using the time dependent sleep stage dynamics which is represented using the hypnogram (see Figure 3)

Methods

Results

Discussion

Conclusion