A Fully Automated Trial Selection Method for Optimization of Motor Imagery Based Brain-Computer Interface

Abstract

Independent component analysis (ICA) as a promising spatial filtering method can separate motor-related independent components (MRICs) from the multichannel electroencephalogram (EEG) signals. However, the unpredictable burst interferences may significantly degrade the performance of ICA-based brain-computer interface (BCI) system. In this study, we proposed a new algorithm frame to address this issue by combining the single-trial-based ICA filter with zero-training classifier. We developed a two-round data selection method to identify automatically the badly corrupted EEG trials in the training set. The “high quality” training trials were utilized to optimize the ICA filter. In addition, we proposed an accuracy-matrix method to locate the artifact data segments within a single trial and investigated which types of artifacts can influence the performance of the ICA-based MIBCIs. Twenty-six EEG datasets of three-class motor imagery were used to validate the proposed methods, and the classification accuracies were compared with that obtained by frequently used common spatial pattern (CSP) spatial filtering algorithm. The experimental results demonstrated that the proposed optimizing strategy could effectively improve the stability, practicality and classification performance of ICA-based MIBCI. The study revealed that rational use of ICA method may be crucial in building a practical ICA-based MIBCI system.

MOABB: trustworthy algorithm benchmarking for BCIs

“…Similarly, while the tangent space projection method normally out-performs CSP, that is also not true for half of the sampled datasets. The confidence intervals also show why this is likely the case -for studies with very few subjects, such as [35], the confidence intervals make even very strong standardized effects quite untrustworthy. Figure 3 compares CSP against commonly used variants.…”

EEG-Based Brain-Computer Interface for Decoding Motor Imagery Tasks within the Same Hand Using Choi-Williams Time-Frequency Distribution

“…The experimental procedure adopted in the current study is similar to the experimental procedures employed in several previous studies related to EEG-based MI tasks classification, such as [ 8 , 34 , 35 , 36 ]. In particular, each subject was seated on a comfortable upright chair at a distance of approximately 0.5 m from a computer monitor placed on top of a desk.…”

“…The Biosemi ActiveTwo system employs the 10–20 international EEG electrode placement system to localize 16 Ag/AgCl electrodes at the following locations: Fp1, Fp2, C3, C4, Cz, F3, F4, Fz, T7, T8, O1, O2, Oz, P3, P4, and Pz, referenced to the common mode sense (CMS)/ driven right leg (DRL) at C1/C2 locations for noise cancelation (see Figure 3 ). In this study, we consider four different groups of electrodes that cover different motor cortex related regions in the brain [ 9 , 36 , 37 ]. Table 2 shows the electrodes included within each group.…”

Dimensionality Transcending: A Method for Merging BCI Datasets With Different Dimensionalities

“…We present examples with BCI data from both MI and P300 paradigms. For MI, we use the Zhou2016 [34], BNCI2015001 [35], and AlexMI [36] datasets. The Zhou2016 dataset consists of recordings on 14 electrodes from 4 subjects executing either a left-hand/righthand or a feet/right-hand motor imagery task; we denote these sub-datasets by Zhou2016-LR (LR for left-hand/righthand) and Zhou2016-FR (FR for feet/right-hand).…”