Wavelet Methodology for EEG Time Frequency (Power Spectrum) Analysis
This literature review examines previous studies that have been performed by different researchers on media communications systems. The primary concern will be to provide a critical analysis of Electroencephalogram (EEG) time-frequency where the key findings on this aspect will be analyzed. Secondly, the literature review will focus on the examination of the concept of wavelet transformations as applied in communication systems. Finally, it will be crucial to identify the limitations in the current research and suggest areas that future studies can focus on while addressing such issues.
EEG Time-Frequency Analysis
The term EEG has varied definitions based on the understanding as well as the discipline of a researcher. Alqazzaz et al. define this term as a neurophysiologic instrument that is utilized in the identification and monitoring of brain signal changes resulting from traumatic brain injury and seizure disorder among other physiological issues (Alqazzaz et al., 2015). The researchers also note that EEG has benefits of being a cost-effective, non-invasive, and broadly available tool. In the context of communication, Bashashati et al. (2007) postulate that the features derived from the signals of EEG concisely represent the distribution of power in distinct bands of frequency. Furthermore, they indicate that the power of EEG is a correspondence of the cortical information processing capacity. Conversely, Li et al. (2017) present EEG as a typical signal that in non-periodic in nature. Thus, this research indicates that it is not possible for the classic approach of analysis to accurately show the characteristic of the local time-varying frequency spectral of the signal.
Schalk and Mellinger (2010) suggest that, to successfully create a new communication channel, which is a direct process from the brain to an output device, any developer relies on two critical aspects. First, there is the utilization of a sufficient sensor, which can measure the features of the brain in an effective manner, aiming at the communication of the intent of the user. Schalk and Mellinger (2010) identify the second requirement as the negotiation and definition of a mutual language for the user to communicate the intent using the symbols of such language.
In the same study, Schalk and Mellinger (2010) point out that the BCI communication language does not bear all the properties of the arbitrary process. One of the reasons for this limitation is that it cannot be possible for the brain to have the physical ability to generate the symbols the language, where its symbols could take the form of the discrete coherent amplitudes. This condition may imply that the brain simply has no physical capability of generating changes in the coherence amplitude at particular locations and frequencies. According to Schalk and Mellinger (2010), despite the brain being able to produce language symbols, it may lack the ability to utilize to communicate the intent. In this case, it is easy for someone to give a definition of arbitrary language as the modulations of the amplitude occurring at 10Hz instead of the visual locations of the brain. In contrast, Schalk and Mellinger (2010) highlight that several studies have indicated that the repetitive visual stimuli taking place at certain frequencies, like 10 Hz, can trigger oscillatory responses in the brain. Therefore, there is a possibility that the brain would modulate the activity at the specified 10 Hz, thereby being able to generate diverse symbols of the arbitrary language.
Wavelet Transformation Analysis
Alqazzaz et al. (2015) consider a wavelet transformation (WT) as a spectral estimation method that is powerful for the analysis of the time-frequency of the signals. Kumar et al. (2008) added that it is a diagnosis technique that is effective and whose introduction was aimed at addressing the issue of signals that are not stationary. One of such signals is EEG, which is an area of interest in this review of the literature. WT can be useful in solving the issues that are associated with the resolution through the division of the data of concern into diverse components of the frequency. It can also be via the evaluation of every component where matching of the resolution to a scale is done. Alqazziz et al. (2015) provide further results that the discrete wavelet transform (DWT) has a little time of computation as compared to the continuous WT. This type of wavelet is not a fast but also a non-redundant transform utilized in the analysis of the high- and low-frequency components in the signals of EEG (Bharkad & Kokare, 2013). The model used to show this scenario involves the processing of DWT through the calculation of the discrete value of the frequency scaling parameters (German-Sallo & Ciufudean, 2012). Moreover, Alqazziz et al. (2015) suggest that it is important to select an appropriate number of levels of decomposition for the analysis of the in the EEG signal by use of DWT. Walters-Williams and Li (2011) concluded that the selection of the number of the levels of decomposition relies on the signals’ dominant frequency as well as the significance of features obtained from the individual components of the wavelet.
On the other hand, Li et al. (2017) consider wavelet transform as a form of a new linear time-frequency analysis approach. They assert that it is involved in making the analysis of multi-scales of the signal by the operation function of translation and dilation among others. This approach constitutes an imperative development of the milestone in the history establishment of reconciling analysis. Li et al. (2017) also argue that the wavelet transformation operates under the principle of Heisenberg uncertainty. However, WT’s window is considered as an adjustable time-frequency window, where a narrow window is applied in the high-frequency region while a wide window is used in low frequency. In connection with this process, WT has been fully embodied in the analysis of multi-resolutions, and this aspect has consistency with the time-varying non-stationary signal characteristic.
Limitations and Recommendation for Further Research
In this study, some limitations have been noted in the literature from different studies. The first limitation that Alqazziz presents is the lack of sufficient data for the analysis and this situation call for a bigger database that will aid effective and reliable analysis in the offing (Alqazziz et al., 2015). There is also little evidence regarding the high gamma bands in the analysis of the current research (Fründ et al., 2007). The reason for this case might have been as a result of the cut-off frequency of the system of EEG recording utilized in previous studies. In this case, some critical information misses and as well the spectral range seems to be noisy because of the muscular artifacts. Previous studies have concentrated on the functions, such as mother wavelet transformation (MWT), which cannot be utilized in all the pathological and physiological conditions of the brain. Therefore, for effective development of the systems of the communication channels, future research should take into account better approaches of EEG signals.
References
Alqazziz, N.K., Ali, S.H.B.M., Ahmad, S.A., Islam, M.S., & Escudero, J. (2015). Selection of Mother Wavelet Functions for Multi-Channel EEG Signal Analysis during a Working Memory Task, Sensor, 15, 29015-29035
Bashashati, A., Fatourechi, M., Ward, R.K., & Birch, G.E. (2007). A survey of signal processing algorithms in brain–computer interfaces based on electrical brain signals. J. Neural Eng. 4(2), R32–R57 (2007)
Bharkad, S., & Kokare, M. (2013, February). Fingerprint matching using discreet wavelet packet transform. In Advance Computing Conference (IACC), 2013 IEEE 3rd International (pp. 1183-1188). IEEE.
Fründ, I., Busch, N. A., Schadow, J., Körner, U., & Herrmann, C. S. (2007). From perception to action: phase-locked gamma oscillations correlate with reaction times in a speeded response task. BMC Neuroscience, 8, 27.
German-Sallo, Z. & Ciufudean, C. (2012). Waveform-adapted wavelet denoising of ECG signals. Advanced Mathematics Computational Methods, 172–175
Kumar, P.S., Arumuganathan, R., Sivakumar, K., Vimal, C. (2008). Removal of ocular artifacts in the EEG through wavelet transform without using an EOG reference channel. International Journal Open Problem Computational Math, 1(1), 188–200.
Li, Y., Zhang, L., Li, B., Xu, Y., Wu, S., Wei, X.,Liu, X., Lin, R., & Wang, Q. (2017). The simulation study of three typical time frequency analysis methods. BIO Web of Conferences, 8.
Schalk, G. & Mellinger, J. (2010). A practical guide to brain–computer interfacing with BCI2000. London: Springer-Verlag London Limited.
Walters-Williams, J. & Li, Y. (2011). Performance comparison of known ICA algorithms to a wavelet-ICA merger. Signal Process. International Journal, 5(1), 80–92.
