Comparison of Sensor Selection Mechanisms for an ERP-Based Brain-Computer Interface

Abstract

A major barrier for a broad applicability of brain-computer interfaces (BCIs) based on electroencephalography (EEG) is the large number of EEG sensor electrodes typically used. The necessity for this results from the fact that the relevant information for the BCI is often spread over the scalp in complex patterns that differ depending on subjects and application scenarios. Recently, a number of methods have been proposed to determine an individual optimal sensor selection. These methods have, however, rarely been compared against each other or against any type of baseline. In this paper, we review several selection approaches and propose one additional selection criterion based on the evaluation of the performance of a BCI system using a reduced set of sensors. We evaluate the methods in the context of a passive BCI system that is designed to detect a P300 event-related potential and compare the performance of the methods against randomly generated sensor constellations. For a realistic estimation of the reduced system's performance we transfer sensor constellations found on one experimental session to a different session for evaluation. We identified notable (and unanticipated) differences among the methods and could demonstrate that the best method in our setup is able to reduce the required number of sensors considerably. Though our application focuses on EEG data, all presented algorithms and evaluation schemes can be transferred to any binary classification task on sensor arrays.

Feature Extraction and Classification Methods for Hybrid fNIRS-EEG Brain-Computer Interfaces

“…Most BCI studies use the international 10–20 (or 10–10) system for placement of EEG electrodes (Homan et al, 1987 ; Jurcak et al, 2007 ). Because EEG responses such as the P300 are highly variable across brain areas, channel selection algorithms are applied to determine a region of interest (ROI) (Feess et al, 2013 ; Alotaiby et al, 2015 ). In fNIRS, channel averaging is used because the hemodynamic responses are not distinctive among channels (Bhutta et al, 2015 ; Naseer and Hong, 2015b ; Liu and Hong, 2017 ).…”

How to evaluate an agent's behavior to infrequent events?—Reliable performance estimation insensitive to class distribution

“…Across research areas different treatments have been proposed for evaluating imbalanced classes such as genetics (Velez et al, 2007; Garcia-Pedrajas et al, 2012), bioinformatics (Levner et al, 2006; Rogers and Ben-Hur, 2009), medical data sets (Cohen et al, 2003, 2004; Li et al, 2010), data mining, and machine learning (Bradley, 1997; Fawcett and Provost, 1997; Kubat et al, 1998; Gu et al, 2008; Powers, 2011). In neuroscience, recent approaches evaluating the performance of brain-computer interfaces are trying to find a more direct and intuitive measure of performance in imbalanced cases (Zhang et al, 2007; Hohne and Tangermann, 2012; Salvaris et al, 2012; Feess et al, 2013). However, the decision for a single metric is often avoided by keeping the numbers for the two classes separated (e.g., Bollon et al, 2009; Kimura et al, 2010).…”

EEG als Steuersignal: Gehirnaktivität entschlüsseln und effizient als Kommunikationsmittel für Patienten mit motorischen Defiziten nutzen

“…Daher wird häufig eine Extraktion von Merkmalen bevorzugt, die es erlaubt, aus den Daten zu lernen, welche Merkmale für die Dekodierung informativ sind. Beispielsweise gibt es verschiedene Ansätze, individuelle Kanäle zu selektieren, die vielversprechend für die Dekodierung sind [2][3][4] [5]) und sich besonders für die Detektion von motorischer Vorstellung bewährt hat, die häufig für die Steuerung von BCIs verwendet wird [6].…”