EEG Channel Interpolation Using Deep Encoder-decoder Netwoks

Abstract

Electrode "pop" artifacts originate from the spontaneous loss of connectivity between a surface and an electrode. Electroencephalography (EEG) uses a dense array of electrodes, hence popped segments are among the most pervasive type of artifact seen during the collection of EEG data. In many cases, the continuity of EEG data is critical for downstream applications (e.g. brain machine interface) and requires that popped segments be accurately interpolated. In this paper we frame the interpolation problem as a self-learning task using a deep encoder-decoder network. We compare our approach against contemporary interpolation methods on a publicly available EEG data set. Our approach exhibited a minimum of ∼ 15% improvement over contemporary approaches when tested on subjects and tasks not used during model training. We demonstrate how our model's performance can be enhanced further on novel subjects and tasks using transfer learning. All code and data associated with this study is open-source to enable ease of extension and practical use. To our knowledge, this work is the first solution to the EEG interpolation problem that uses deep learning. I. INTRODUCTIONElectroencephalography (EEG) devices have become increasingly popular in recent years and are used in a wide range of applications. Naturally, the medical applications of EEG are centered on neurological diagnosis, but EEG has proven useful for other problems in healthcare domain [1], [2]. Moreover, the use of EEG devices extends far beyond the medical domain; novel applications of EEG may be found in wide a variety of fields including advertising [3], education[4], entertainment [5], and security [6].A fundamental challenge of EEG data is the low signal to noise ratio. Different sources contribute to this noisiness but, in general, they can be categorized as either movement artifacts or electrode artifacts. The most common, and particularly persistent, electrode artifact is the electrode "pop" [7], [8]. These artifacts result from abrupt changes in impedance, usually due to a loose electrode or bad conductivity. Furthermore, these artifacts are difficult to avoid because, even if the greatest care is taken when applying electrodes, the most minor subject movement or change in perspiration can cause the electrode to "pop".A common solution to EEG "pops" is to interpolate the missing segments using recordings from nearby electrodes

Reconstruction of missing channel in electroencephalogram using spatiotemporal correlation-based averaging

“…The proposed weighting was inspired by the general expression used for the weighted average formula in the literature [13,14,18,[41][42][43].…”

“…So far, there have been a few studies investigating the reconstruction of missing channels in physiological time series [14][15][16][17][18][19][20]. These can be categorized into two major groups of simple and complex techniques.…”