Fig 1.
Illustration of a subject participating in the experimental procedure.
Fig 2.
The experimental setup used for simultaneous EEG and NIRS signal acquisition.
Fig 3.
A schematic diagram of the experimental paradigm illustrating the sequence and structure of the experiment.
Fig 4.
The visual screen displaying the scrolling text used during the experiment.
The text flows in one of four directions-right, left, up or down-depending on the task instructions given at the beginning of each trial.
Fig 5.
Illustration of EEG electrode positions next to NIRS emitters and detectors.
The EEG electrodes, NIRS detectors and NIRS emitters are represented by blue, purple and green circles, respectively. NIRS channels are represented by black solid lines.
Fig 6.
General flow chart of the proposed method.
Table 1.
Confusion matrix for multiclass classification.
Fig 7.
Display of subject-specific bar graphs showing average CA as percentages across different modalities.
Fig 8.
Average CA Results for 2.4-Second Time Segments: a) Training Set and b) Test Set, representing both single and hybrid modalities in terms of percentage.
Fig 9.
Average CA Results for 4.8-Second Time Segments: a) Training Set and b) Test Set, representing both single and hybrid modalities in terms of percentage.
Fig 10.
Average CA Results for 12-Second Time Segments: a) Training Set and b) Test Set, representing both single and hybrid modalities in terms of percentage.
Fig 11.
Average CA Results for 24-Second Time Segments: a) Training Set and b) Test Set, representing both single and hybrid modalities in terms of percentage.
Table 2.
Average CA for HT, FWHT, BP, and SF-based features extracted from 2.4-second time segments using the k-NN classifier.
Table 3.
Average ca for HT-based features extracted from 2.4-second time segments using classification methods.
Table 4.
The computational time of feature extraction for a single trial across different time segments.
Table 5.
Studies in the literature are compared to the proposed method.