For diagnostic and treatment purposes brain activity in comatose patients must be accurately classified. Existing classification method use Support Vector Machines that often cause problems in processing high-dimensional and complex brain signal data. On the other hand this study proposes a new method by integrating Convolutional Neural Network(CNN)- based analysis with features of Wavelet Transform(WT) and Continuous Stock well Ttransform(CST). While CST functions to capture spatiotemporal dynamics improving the representation of brain activity, WT effectively divides brain signals into multiple frequency bands and CNN model is used to classify brain activity and detect complex patterns. Experimental results show that our proposed model performs much better in classification than SVM-based techniques. This method appears promising for real-time monitoring of comatose patients and may be extended to other neurological disorders. We anticipate upcoming initiatives that will attempt to improve the integration of the model into the clinical setting.
Keywords
Brain Activity Classification, Coma Patients, Wavelet Transform, Continuous stock well Transform, Convolutional Neural Network, Support Vector Networks.
Conclusion
This study deployed a Convolutional Neural Network
to analyze brain activity and create an EEG categorization framework. Discrete Wavelet
Transform and Continuous Stockwell Transform were used to extract features from the EEG data after they had been pre-processed using a band pass filter to eliminate unwanted noise. Subsequently, these characteristics were employed to categorize brain states and differentiate between various activity levels. Gaussian noise was included to the collected features to increase resilience and decrease overfitting, and a more difficult data split was used to improve model generalization.
The CNN model is a feasible method for EEG data analysis since it showed good feature learning and classification accuracy. Reliable classification was supported by the network’s ability to recognize complex patterns in electroencephalography (EEG) data thanks to the inclusion of convolutional layers… We can certainly explore further development using various deep learning models. Like For example long-term, memory networks are an excellent choice because they handle sequential EEG data much better. Feature selection, methods such as principal component analysis can be incorporated to maximize feature representation. In order to improve generalizability and performance in practical EEG- based applications, future study may potentially entail adjusting hyperparameters and evaluating the model on bigger datasets.
References
A. Bhardwaj,, . A. Tiwari,, . D. Gupta,, ,. P. Tripathi, Jhanjhi and . J. N. Z., , “”A novel deep learning framework for classification of EEG
brain signals”,” Applied Sciences, Vols. 11(17),, p. 8026, 2018.
[2] L. Y, . S. O and . N. M., “Real-time EEG brainwave monitoring for mental state recognition,” Proc. Int. Conf. Cyber World, pp. 153-159, 2009.
[3] N. E and d. S. . L. F., “Electroencephalography: Basic Principles, Clinical Applications, and Related Fields,” Lippincott Williams & Wilkins, vol. 5th, 2004.
[4] D. . L. L. , L. L. P. and . W. . F. Y., “Brain-
computer interface-based stroke rehabilitation,,” ” IEEE Trans. Biomed. Eng, vol. 59, pp. 2836- 2844, 2012.
[5] . H. H. and . W. D, “Transfer learning for brain- computer interfaces: A Euclidean space data alignment approach,” IEEE Trans. Biomed. Eng, vol. 66, pp. 424-432, 2019.
[6] . Z. Y., .. L. T and . S. J., “An fMRI-based approach for brain state classification in coma patients,” Neuroimaging, vol. 118, pp. 124-131,
2015.
[7] Schirrmeister, . T. . S. R., T. F. J. , D. G. L. , E. M., and T. K., , “”Deep learning with convolutional neural networks for EEG decoding and visualization,” Human Brain Mapping, Vols. 38(11),, pp. 5391-5420..
[8] ,. I. Ullah, ,. M. Hussain, E. U. Qazi, and Aboalsamh, “An automated system for epilepsy detection using EEG brain signals based on deep learning approach,” Expert Systems with Applications, vol. 107, pp. 61-71, (2018).
[9] Bashar and S. K, “Deep learning for EEG-based brain activity classification A review,” Biomedical Signal Processing and Control, p. 104172, 2023.
[10] A. . H. Craik and . J. . L. Contreras-Vidal,, “Deep learning for electroencephalogram (EEG) classification tasks: A review,” Journal of Neural Engineering, vol. 16(3), p. 031001,
2015.
[11] S. e. a. Fazli, “Enhanced performance by hybrid EEG-NIRS brain-computer interface,” NeuroImage, vol. 49(1), p. 519–532, 2009.
[12] X. e. a. Gao, “Deep learning for EEG-based brain-computer interfaces: A review,” IEEE Transactions on Neural Networks and Learning Systems, vol. 33(4), p. 1449–1465, 2022.
[13] Z. e. a. Guo, “EEG-based motor imagery classification using a deep convolutional neural
network with efficient kernel normalization,”
Sensors, vol. 20(5), p. 1288, 2020.
H. e. a. He, “Wavelet transform and CNN-based EEG classification for brain state monitoring,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 29, p. 1205–
1216., 2021.
H. e. a. Yang, “Brain activity classification using wavelet transform and CNN for coma patient monitoring,” IEEE Access, vol. 9, p. 12845– 12858, 2021.
R. e. a. Sharma, “Wavelet-based EEG feature extraction for coma state classification using deep learning,” Biomedical Signal Processing and Control, vol. 86, p. 104938, (2023).
J. e. a. Stender, “Diagnostic precision of PET imaging and functional MRI for detecting consciousness in vegetative and minimally conscious states,” The Lancet, vol. 384(9942), p. 514–522, 2014.
S. e. a. Mitra, “”Wavelet-based EEG feature extraction and deep learning for brain activity classification in comatose patients,” IEEE Journal of Biomedical and Health Informatics, vol. 26(8), p. 3742–3751, 2022.
J. W. Y. e. a. Kam, “Brain state classification using deep learning on EEG signals,” Journal of Neuroscience Methods, vol. 328, p. 108387,
2019.
J. e. a. Pan, “Motor imagery brain-computer interface using common spatial patterns and convolutional neural networks,” Journal of Neural Engineering, vol. 15(6), p. 066017,
2018.
How to Cite This Paper
Gagan S, Dr Prabhavathi S (2026). Multidomain Wavelet–CST Representation for Deep Learning Classification of Coma Brain Signals. International Journal of Computer Techniques, 13(2). ISSN: 2394-2231.
