Decoding Imagined Speech: A Strictly Subject-Independent Approach Using EEG

2026-09-24

AI Summary

This study addresses the challenges of insufficient cross-subject generalization and opaque evaluation protocols in imagined speech EEG decoding. Within a strictly subject-independent framework, we compare time-domain statistical and frequency-domain spectral power feature extraction pipelines, employing random forests with forward feature selection for multi-class decoding. This work establishes a transparent baseline evaluation paradigm and demonstrates that a limited subset of frequency bands suffices to capture core discriminative information, thereby significantly enhancing cross-subject generalizability. Experimental results indicate that the frequency-domain pipeline achieves a substantially higher accuracy (49.03%) compared to the time-domain approach (37.97%). These findings provide robust empirical evidence for improving the cross-subject performance of brain-computer interface systems.

Abstract

Imagined speech decoding from electroencephalography (EEG) has gained increasing attention as a potential communication pathway for individuals with severe motor impairments, yet reported performance often relies on evaluation protocols that do not clearly reflect cross-subject generalization. This study presents a transparent baseline investigation of a multi-class imagined speech EEG dataset under a strictly subject-independent evaluation framework. Two preprocessing and feature extraction pipelines were compared: a time-domain statistical feature approach and a frequency-domain spectral bandpower approach, evaluated using subject-wise cross-validation and trial-level majority voting with a random forest classifier. The spectral pipeline achieved a significantly higher mean trial-wise accuracy than the statistical pipeline (49.03 $\pm$ 4.18% vs. 37.97 $\pm$ 3.79%) for coarse-level classification across subjects. Forward feature selection further indicated that a limited subset of frequency bands captured most of the discriminative information. Overall, this work provides a strong basis for future brain-computer interface studies targeting improved cross-subject generalization in EEG-based imagined speech decoding.

Problem

• Imagined Speech Decoding
• EEG
• Subject-Independent
• Cross-subject Generalization
• Brain-Computer Interface

Innovation

• Imagined Speech Decoding
• Subject-Independent
• EEG
• Spectral Bandpower
• Brain-Computer Interface