This systematic review examines EEG-based imagined speech classification, emphasizing directional words essential for development in the brain–computer interface (BCI). This study employed a structured methodology to analyze approaches using public datasets, ensuring systematic evaluation and validation of results. This review highlights the feature extraction techniques that are pivotal to classification performance. These include deep learning, adaptive optimization, and frequency-specific decomposition, which enhance accuracy and robustness. Classification methods were explored by comparing traditional machine learning with deep learning and emphasizing the role of brain lateralization in imagined speech for effective recognition and classification. This study discusses the challenges of generalizability and scalability in imagined speech recognition, focusing on subject-independent approaches and multiclass scalability. Performance benchmarking across various datasets and methodologies revealed varied classification accuracies, reflecting the complexity and variability of EEG signals. This review concludes that challenges remain despite progress, particularly in classifying directional words. Future research directions include improved signal processing techniques, advanced neural network architectures, and more personalized, adaptive BCI systems. This review is critical for future efforts to develop practical communication tools for individuals with speech and motor impairments using EEG-based BCIs.

Classifying imagined speech from electroencephalographic signals remains a significant challenge in brain–computer interface research due to the low amplitude, nonstationary properties, and high intersubject variability of the neural activity involved. However, a key unresolved problem is the poor generalizability of subject-independent imagined-speech EEG classification, largely due to inadequate multiscale signal representations and suboptimal integration of feature extraction, selection, and classification. This study introduces a subject-independent classification framework that integrates multiscale signal decomposition, statistical feature selection, and ensemble classification. Specifically, Rational-Dilation Wavelet Transform is employed to achieve flexible time-frequency analysis, followed by the extraction of time-domain, energy-based, and fractal features from each subband. Statistically discriminative features are identified using one-way analysis of variance, and classification is performed using an Extreme Gradient Boosting model with automated hyperparameter tuning. Evaluated on the Thinking Out Loud dataset, a publicly available imagined speech corpus, the proposed pipeline achieved a subject-independent test set accuracy of 64.45%, substantially outperforming prior methods that typically reported accuracies between 25.8% and 33.9%. The cross-validation results closely matched the performance of the test set, indicating a strong generalization. These findings demonstrate the effectiveness of signal-aware decomposition and statistically principled feature engineering in improving the robustness of imagined speech classification. The proposed approach provides a promising basis for future developments in noninvasive neural decoding and silent communication BCIs.