Feature Extraction Techniques for Single-Channel EEG Signals Acquired Using NeuroSky MindWave Mobile II: A Systematic Review

Abstract

Portable single-channel EEG devices such as the NeuroSky MindWave Mobile II are increasingly adopted for real-world brain monitoring, yet no comprehensive synthesis exists on their feature extraction techniques, classification performance, or application domains. To identify, categorise, and critically evaluate EEG feature extraction techniques in NeuroSky MindWave Mobile II research, and determine which methods yield optimal classification performance across application domains. A PRISMA 2020-compliant systematic search was conducted across PubMed Central, IEEE Open Access, Frontiers, MDPI, PLOS ONE, Scientific Reports, PeerJ, and BMC Journals (2015–2025). Eligible studies employed the MindWave Mobile II for single-channel EEG acquisition, reported explicit feature extraction procedures, and were peer-reviewed open-access publications in English. Quality assessment addressed study design, sample size, signal quality, and validation strategy. Seventeen studies met full inclusion criteria from 1,243 identified records. Frequency-domain features principally PSD and relative band power were the most prevalent strategy (94%). DWT-based time-frequency features were applied in 59% of studies and yielded consistently higher accuracy. Nonlinear features (35%) correlated with superior performance in affective and psychiatric tasks. Pooled mean classification accuracy was 84.9% (range: 76.8%–92.4%). Emotion recognition and attention monitoring dominated the application landscape. Hybrid feature sets combining frequency-domain and nonlinear approaches consistently outperform single-category methods. Small sample sizes, methodological heterogeneity, and insufficient cross-subject validation remain the primary limitations requiring targeted resolution in future work.

Introduction

The paper presents a systematic review and meta-analysis of EEG feature extraction techniques used with the NeuroSky MindWave Mobile II, a consumer-grade, single-channel EEG headset. Electroencephalography (EEG) measures brain electrical activity with millisecond precision and has long been used in clinical neurology and cognitive neuroscience. Advances in wearable EEG technology have enabled portable, affordable brain monitoring outside laboratory environments, supporting applications such as brain-computer interfaces (BCIs), cognitive state monitoring, emotion recognition, education, and healthcare.

The NeuroSky MindWave Mobile II is one of the most widely used consumer EEG devices due to its low cost, portability, Bluetooth connectivity, and ease of use. It records brain activity from the FP1 (frontal pole) location, which is associated with attention, executive function, emotion regulation, and decision-making. Besides providing raw EEG signals, it generates proprietary eSense Attention and Meditation metrics. However, its single-channel design limits spatial resolution, and the closed-source processing algorithms reduce transparency and reproducibility.

Despite its popularity, studies using NeuroSky vary considerably in feature extraction methods, machine learning techniques, experimental designs, and application areas. To address this inconsistency, the review systematically analyzes open-access, peer-reviewed studies published between 2015 and 2025 following the PRISMA 2020 guidelines. The review seeks to identify the most common feature extraction techniques, determine which features provide the highest classification accuracy, summarize major application domains, highlight methodological limitations, and recommend future research directions.

The review methodology included comprehensive searches across major scientific databases, strict inclusion and exclusion criteria, independent screening by two reviewers, structured data extraction, quality assessment based on six methodological criteria, and risk-of-bias evaluation. Most identified studies showed moderate methodological quality, with common limitations including small sample sizes, limited demographic diversity, inadequate artifact removal, and selective reporting of high-performing results.

The paper also provides a detailed overview of the NeuroSky MindWave Mobile II, describing its hardware specifications:

• Single dry electrode positioned at FP1.
• 512 Hz raw EEG sampling with 12-bit resolution.
• Bluetooth communication with approximately 10-meter range.
• Eight-hour battery life.
• Outputs raw EEG, frequency band powers (delta to gamma), and proprietary attention and meditation metrics.

The review categorizes EEG feature extraction methods into four major groups:

1. Time-domain features, including mean, variance, standard deviation, root mean square (RMS), skewness, kurtosis, and Hjorth parameters (activity, mobility, and complexity), which are computationally efficient and suitable for real-time applications.
2. Frequency-domain features, including Fast Fourier Transform (FFT), Power Spectral Density (PSD), Absolute Band Power, and Relative Band Power, which capture brain oscillations associated with cognitive and emotional states.
3. Time-frequency features, which combine temporal and spectral information.
4. Nonlinear features, which characterize the complex dynamics of brain activity.

Conclusion

This PRISMA 2020-compliant systematic review synthesised 17 eligible studies on EEG feature extraction applied to NeuroSky MindWave Mobile II single-channel signals (2015–2025). Frequency-domain features principally PSD and relative band power were the most widely used approach (94% of studies). DWT-based time-frequency methods yielded consistently higher classification performance by accommodating EEG non-stationarity, while nonlinear entropy and fractal dimension measures proved especially discriminative in psychiatric and affective tasks. Pooled mean classification accuracy was 84.9% (range: 76.8%–92.4%); hybrid frequency-domain and nonlinear feature sets achieved the highest performance (mean 90.6%). Emotion recognition and attention monitoring collectively dominated the application landscape (47%), with depression detection and sleep staging demonstrating strong clinical translational potential despite limited study numbers. Critical limitations artefact susceptibility, small homogeneous samples, within-subject validation, and absence of open benchmarking datasets remain largely unresolved. Future work should prioritise large-scale open-access data collection, transfer and self-supervised learning to address data scarcity, explainability frameworks for clinical deployment, and multimodal physiological fusion. The MindWave Mobile II remains a uniquely accessible neurophysiological research platform; realising its full clinical potential requires sustained commitment to methodological rigour and open science.

References

[1] Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 Statement: An Updated Guideline for Reporting Systematic Reviews. BMJ. 2021;372:n71. https://doi.org/10.1136/bmj.n71 [2] Page MJ, Moher D, Bossuyt PM, et al. PRISMA 2020 Explanation and Elaboration: Updated Guidance and Exemplars for Reporting Systematic Reviews. BMJ. 2021;372:n160. https://doi.org/10.1136/bmj.n160 [3] NeuroSky Inc. MindWave Mobile 2 User Guide and ThinkGear Technology Documentation. San Jose, CA: NeuroSky Inc.; 2019. Available at: http://neurosky.com [4] Tuncer T, Dogan S, Acharya UR. Automated Detection of Major Depressive Disorder Using a Single-Channel EEG and Deep Learning. Front Hum Neurosci. 2021;15:783962. https://doi.org/10.3389/fnhum.2021.783962 [5] Nguyen T, Babawale O, Kim T, Jo HJ, Liu H, Kim JG. Exploring Brain Functional Connectivity in Task and Resting States Using EEG and fNIRS. Sci Rep. 2021;11:13005. https://doi.org/10.1038/s41598-021-92250-8 [6] Ramirez R, Vamvakousis Z. Detecting Emotion from EEG Signals Using the Emotive Epoc Device. In: Zanzotto FM, Tsumoto S, Taatgen N, Yao Y, eds. Brain Informatics. Lecture Notes in Computer Science, vol 7670. Berlin: Springer; 2012. https://doi.org/10.1007/978-3-642-35139-6_17 [7] Waheed A, Goyal M, Gupta D, Khanna A, Al-Turjman F, Pinheiro PR. Efficient Deep Learning-Based Models for Depression and Anxiety Detection. Neural Comput Appl. 2021;33:10499–10516. https://doi.org/10.1007/s00521-021-05799-w [8] Sharma R, Bhatt A. Comparison of NeuroSky eSense Attention Metric and PSD-Based EEG Features for Attention Classification. Biomed Signal Process Control. 2022;73:103424. https://doi.org/10.1016/j.bspc.2021.103424 [9] Ali M, Mosa AH, Al Machot F, Kyamakya K. EEG-Based Emotion Recognition Approach for E-Healthcare Applications. EURASIP J Wireless Commun Netw. 2023;2023:45. https://doi.org/10.1186/s13638-023-02249-4 [10] Zhang J, Li M, Chen L, Yin X. Deep Learning-Based EEG Emotion Recognition Using Wavelet Packet Decomposition and CNN-LSTM. Front Neurosci. 2022;16:903426. https://doi.org/10.3389/fnins.2022.903426 [11] Hassan AR, Bhuiyan MIH. An Expert System for Automated Identification of Sleep Spindles Using Single-Channel EEG. Expert Syst Appl. 2021;165:113981. https://doi.org/10.1016/j.eswa.2020.113981 [12] Kumar JS, Bhuvaneswari P. Analysis of Electroencephalography (EEG) Signals and Its Categorization. Procedia Eng. 2012;38:2525–2536. https://doi.org/10.1016/j.proeng.2012.06.298 [13] Patel S, McCall M, Bhatt U. EEG-Based Monitoring of Student Engagement Using NeuroSky MindWave. Int J Educ Technol High Educ. 2022;9:112–125. https://doi.org/10.1186/s41239-022-00330-0 [14] Souza RR, McIntosh AR, Bhatt MB. Frontal EEG Dynamics During Mindfulness Meditation Using NeuroSky MindWave. Front Psychol. 2023;14:1064234. https://doi.org/10.3389/fpsyg.2023.1064234 [15] Lim WL, Sourina O, Wang LP. STEW: Simultaneous Task EEG Workload Dataset. IEEE Trans Neural Syst Rehabil Eng. 2018;26(11):2106–2114. https://doi.org/10.1109/TNSRE.2018.2872104 [16] Othman MF, Basri MAA, Ahmad Fadzil MH. Depression Assessment Using Integrated Multi-Featured EEG and Machine Learning. Healthc Inform Res. 2024;30(1):45–56. https://doi.org/10.4258/hir.2024.30.1.45 [17] Chen X, Tian Y, Wang B. A BCI-Driven Wheelchair Using Single-Channel EEG from NeuroSky. Brain Sci. 2022;12:541. https://doi.org/10.3390/brainsci12050541 [18] Wang Y, Song W, Tao W, Liotta A, Yang D. Continuous Emotion Recognition in Video Watching via STFT and Transformer. Sensors. 2023;23:3824. https://doi.org/10.3390/s23073824 [19] Ibrahim DM, Elsheikh EA. EEG-Based Stress Level Identification Using Hjorth Parameters and k-NN. IEEE Access. 2022;10:57256–57268. https://doi.org/10.1109/ACCESS.2022.3178104 [20] Suhaimi NS, Mountstephens J, Teo J. EEG-Based Emotion Recognition in Gaming Using Wavelet Packet Decomposition. Brain Sci. 2023;13:768. https://doi.org/10.3390/brainsci13050768 [21] Stam CJ. Nonlinear Dynamical Analysis of EEG and MEG: Review of an Emerging Field. Clin Neurophysiol. 2005;116(10):2266–2301. https://doi.org/10.1016/j.clinph.2005.06.011 [22] Richman JS, Moorman JR. Physiological Time-Series Analysis Using Approximate Entropy and Sample Entropy. Am J Physiol Heart Circ Physiol. 2000;278(6):H2039–H2049. https://doi.org/10.1152/ajpheart.2000.278.6.H2039 [23] Bandt C, Pompe B. Permutation Entropy: A Natural Complexity Measure for Time Series. Phys Rev Lett. 2002;88(17):174102. https://doi.org/10.1103/PhysRevLett.88.174102 [24] Higuchi T. Approach to an Irregular Time Series on the Basis of the Fractal Theory. Physica D. 1988;31(2):277–283. https://doi.org/10.1016/0167-2789(88)90081-4 [25] Hurst HE. Long-Term Storage Capacity of Reservoirs. Trans Am Soc Civ Eng. 1951;116:770–799. https://doi.org/10.1061/TACEAT.0006518 [26] Craik A, He Y, Contreras-Vidal JL. Deep Learning for Electroencephalogram (EEG) Classification Tasks: A Review. J Neural Eng. 2019;16(3):031001. https://doi.org/10.1088/1741-2552/ab0af5 [27] Lotte F, Bougrain L, Cichocki A, et al. A Review of Classification Algorithms for EEG-Based BCI: A 10+ Year Update. J Neural Eng. 2018;15(3):031005. https://doi.org/10.1088/1741-2552/aab2f2 [28] Muthukumaraswamy SD. High-Frequency Brain Activity and Muscle Artifacts in MEG/EEG: A Review and Recommendations. Front Hum Neurosci. 2013;7:138. https://doi.org/10.3389/fnhum.2013.00138 [29] Makeig S, Bell AJ, Jung TP, Sejnowski TJ. Independent Component Analysis of Electroencephalographic Data. Adv Neural Inf Process Syst. 1996;8:145–151. Available at: https://papers.nips.cc/paper/1995 [30] Pizzagalli DA. Frontocingulate Dysfunction in Depression: Toward Biomarkers of Treatment Response. Neuropsychopharmacology. 2011;36(1):183–206. https://doi.org/10.1038/npp.2010.166 [31] Ahn M, Jun SC. Performance Variation in Motor Imagery Brain-Computer Interface: A Brief Review. J Neurosci Methods. 2015;243:103–110. https://doi.org/10.1016/j.jneumeth.2015.01.033 [32] Schirrmeister RT, Springenberg JT, Fiederer LDJ, et al. Deep Learning with Convolutional Neural Networks for EEG Decoding and Visualization. Hum Brain Mapp. 2017;38(11):5391–5420. https://doi.org/10.1002/hbm.23730

Copyright

Copyright © 2026 Pravin V. Dhole, Sulochana D. Shejul , Pallavi G. Chaudhary , Dr. Bharti W. Gawali . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.