Uniform approximation by spherical spline interpolation

BACKGROUND Electroencephalography (EEG) is a non-invasive Brain Computer Interface (BCI) technology that has shown potential in various healthcare applications such as epilepsy treatment, sleep disorder diagnosis, and stroke rehabilitation. Usually these applications require multi-channels EEG. However, multi-channel EEG headset setup process is time consuming. This may result in low patients’ acceptance despite BCI potential benefits. OBJECTIVE To investigate the number of appropriate electrodes, which could be crucial for successful applications of BCI in wearable devices. METHODS Motor Imagery (MI) classification system is used for our analysis. Different number of EEG channels was selected. EEG Multi-frequency features were extracted by Filter Bank (FB). Support Vector Machine (SVM) was used in classifying left and right hand opening/closing MI task. RESULTS The results showed that the group of nine electrodes gave high classification accuracy while requiring moderate set-up time, and hence is suggested as the minimal number of channels. Spherical spline interpolation (SSI) was also applied to investigate the feasibility of generating EEG signal from limited channels of EEG headset. The classification accuracies of the interpolated groups only, and the combined interpolated and collected group, were significantly lower than those of measured groups CONCLUSIONS For wearable device, one of the key factors that need to be concerned is wearability. The number of channels of EEG device adversely affects to set-up time. With FB feature and session dependent training, the investigation of number of channels provides the possibility to develop a successful BCI application using minimal channels EEG device. Interpolation technique which could approximate additional electrode data from nearby electrodes should be also explored.

Download Full-text

Virtual EEG-electrodes: Convolutional neural networks as a method for upsampling or restoring channels

10.1101/2020.04.20.049916 ◽

2020 ◽

Author(s):

Mats Svantesson ◽

Håkan Olausson ◽

Anders Eklund ◽

Magnus Thordstein

Keyword(s):

Neural Networks ◽

Network Performance ◽

Spline Interpolation ◽

University Hospital ◽

Practical Reasons ◽

Visual Evaluation ◽

Temple University ◽

Statistical Measures ◽

Spherical Spline ◽

The Temple

ABSTRACTBackgroundIn clinical practice, EEGs are assessed visually. For practical reasons, recordings often need to be performed with a reduced number of electrodes and artifacts make assessment difficult. To circumvent these obstacles, different interpolation techniques can be utilized. These techniques usually perform better for higher electrode densities and values interpolated at areas far from electrodes can be unreliable. Using a method that learns the statistical distribution of the cortical electrical fields and predicts values may yield better results.New MethodGenerative networks based on convolutional layers were trained to upsample from 4 or 14 channels or to dynamically restore single missing channels to recreate 21 channel EEGs. 5,144 hours of data from 1,385 subjects of the Temple University Hospital EEG database were used for training and evaluating the networks.Comparison with Existing MethodThe results were compared to spherical spline interpolation. Several statistical measures were used as well as a visual evaluation by board certified clinical neurophysiologists. Overall, the generative networks performed significantly better. There was no difference between real and network generated data in the number of examples assessed as artificial by experienced EEG interpreters whereas for data generated by interpolation, the number was significantly higher. In addition, network performance improved with increasing number of included subjects, with the greatest effect seen in the range 5 – 100 subjects.ConclusionsUsing neural networks to restore or upsample EEG signals is a viable alternative to interpolation methods.

Download Full-text