Google
 
21st Home / Contact Us /Archives

 

 

 

On The Methodology Of EEG Analysis
During Altered States Of Consciousness

Emil Jovanov

e-mail: jovanov@ieee.org

Electrical and Computer Engineering Dept.
The University of Alabama in Huntsville
Huntsville, AL 35899


This is an exciting study on brainwaves and healing, and is also from the groundbreaking new book, "CONSCIOUSNESS, SCIENTIFIC CHALLENGE OF THE 21ST CENTURY," published by the United Nations.

If the doors of perception were cleansed every thing would
appear to man as it is, infinite. -- William Blake

Abstract

Altered states of consciousness exhibit subtle EEG changes that must be observed with carefully chosen methodology and signal processing procedures. Generalized psycho-physiological model of the Self is introduced to point out the possible sources of nervous system excitation that are particularly important in altered states of consciousness. The proposed methodology is based on combination of static (artifact-free EEG) and dynamic analysis to characterize underlying neurophysiological states. We developed open software environment STATE (Spatio-Temporal EEG Alteration Tracing Environment) to support this methodology and provide support for signal processing functions and efficient visualization procedures. Proposed methodology and software environment are used for analysis of brain activities during altered state of consciousness related to the healing process. We present here obtained results and new parameters in quantitative EEG analysis that can be efficiently used to characterize state of consciousness.

Introduction

At the end of 20th century, contemporary science considers itself capable to cope with the ultimate secret of the Nature - consciousness. Philosophers, psychologists, neurophysiologysts, physicist, engineers and other scientists investigate the problem from their own point of view, like shadows on their wall of Plato's cave, but the answer must be one and unique, as the consciousness itself. How can we use EEG, as a crude measure of temporal activity of 1010 neurons, to study such a subtle phenomenon? To paraphrase gestalt psychologist Kurt Goldstein, if consciousness appears on points of contact between organism and environment, then EEG and MEG might be the tools of choice.

Relation between EEG and gross neurophysiological changes in states of consciousness (alertness vs. sleep, sleep phases, coma, epileptic seizures, etc.) is well established and analyzed [1-4]. The situation is somewhat vague for subtle changes, but we believe that every subtle change generates evident physical equivalent (and vice versa). Altered states of consciousness as extreme cases are indispensable in studying nature of consciousness.

Successful neural network models of parts of neural system inspired so called "connectionists" approach that led to the conclusion that consciousness is "in fact no more than behavior of a vast assembly of nerve cells and their associated molecules" [5]. However, the largest problem with this approach is still explanation how brain integrates fragments of information derived from highly specialized set of neurons to create unity of perception and thought. This problem is called binding problem. The above approach can hardly explain how can we perceive gestalts from a large amount of sensory information almost instantly.

The second approach is based on neural fields. In this model, the electromagnetic field of brain activity binds together particular parts of information [6-10]. Brain wave patterns therefore represent internal language of the brain and create local resonance (see Adey, John, Basar, and Leinfellner in [4]). Strong support for this approach is derived from the theory of coupled oscillators and spontaneous synchronization of biological systems [11]. Deterministic chaos is frequently used in explaining brain dynamics through the last decade. Moreover, chaotic systems are capable of producing novel activity patterns. That feature may influence brain's creativity and trial-and-error problem solving.

We believe that certain pattern(s) of EM field activity represent basis from which different states of consciousness arise. Our hypothesis is that those patterns, although subtle, could be detected in brain electrical activity. In this paper, we present framework of analysis that could be used to characterize subtle EEG changes.

Generic model

Although it was not perfectly clear what consciousness is, we will try to use engineering approach to the problem using "black box" model of the Self as a conscious entity. Figure 1 depicts generalized psycho-physiological model of the Self. It consists of three major blocks:

  • Perception is input block, processing sensory inputs (vision, sound, touch, ...).

  • Action generates different outputs (thoughts, emotions, physical actions, ...)

  • Conscious processing block interacts with both blocks and influences their activity according to the genetic heritage and the history of processing (sequence of inputs and action feedback representing experience). Some of its functions are simple connections between neurons (reflexes for example), and some emerge as complex interactions.

In spite of the fact that model of such simplicity must have high complexity elements (in this case conscious processing block), it can point out very important aspects of system functionality.

 

Figure 1: Generalized psycho-physiological model of the Self

Extreme states of consciousness are mostly related to the modified functionality of Perception block. Its output is changed by overloading senses, sensory deprivation or by changing its functionality (drug admission for example) [12]. Having in mind that during these states Action block exhibits rather altered output than its cessation, we can draw a conclusion that our model must have either internal generator or set of inputs that is not dependent on senses. We will introduce both possibilities in our model:

  1. Internal signal generators within the Conscious processing block, represented as a white circle.

  2. Extrasensory inputs as a means of communication with environment. They are represented as a set of separated inputs coming to the shaded part of Perception block.

Internal signal generators are set of physiological control loops within the organism, such as heart or breath control loop. Their fields influence significantly course of conscious processing, but their function is also influenced by the overall conscious state. The breath control loop is exceptionally important as it can be easily consciously controlled. It was shown that a mixture of combined yogic practices of breathing and relaxation (Santhi Kriya) increased alpha activity both in occipital and pre-frontal areas of the brain denoting an increase of calmness [13].

Extrasensory input may help in explaining most of psychic phenomena [14]. If electromagnetic (EM) field plays crucial role in explaining higher functions of conscious processing, then inter-personal influence of particular EM field must be taken into account. One can argue that the intensity of such field is negligible to induce any action. However, we should have in mind that basic processes within the brain are based on resonance [10], and that even slow intensity field may "provoke" resonant patterns of activity [15]. Framework for this analysis is given through the theoretical model of Dejan Rakovi. [16-19].

Throughout our research we are looking for the scientific evidence of both generators in our model, and characterization of their influence on consciousness. We believe that answers to these questions may offer clues for understanding very nature of consciousness.

3. EEG changes during altered state of consciousness

According to the generic model and framework of our investigation we will pay our attention particularly on conscious change of state of consciousness and inter-personal interactions. The most prominent examples of the first kind are meditation, relaxation, and similar techniques, and the healing process presents the most significant example of the second kind. We investigated inter-personal interactions through the healing process, as a particularly intense conscious effort to establish inter-personal communication and help the patient [20-21].

Most of research associated with the changes in brain electrical activity (BEA) in altered state of consciousness are related to meditation [22-29]. The most important features of EEG changes related to meditation are:

  1. Establishing alpha activity in spite of open eyes (Hirai [22])

  2. Increased amplitude of alpha activity (Hirai [22], Banquet [23,24], Wallace [25])

  3. Slower frequency of alpha rhythm (Hirai [22], Banquet [23,24], Wallace [25]

  4. Rhythmical theta waves (Hirai [22], Banquet [23,24], Wallace [25]

  5. Increased synchronization (hypersynchronization - Banquet [23,24])

  6. Dissociation of perception from the external sense organs (Hirai [22], Ray [27])

  7. Transcendent signal (Ray [27,28])

  8. Occasional fast wave activity (Banquet [24], Das and Gastaut [25], and Ray [27])

The first four changes are reported during the study of EEG changes related to Zen meditation [22]. Kasamatsu and Hirai ranked the changes in this order and find out that the changes directly depend on mental state and experience in meditation. During zazen (Zen meditation) alpha was slowing to 7-8 Hz, and rhythmical theta waves at six to seven cycles per second appeared in the last phase (attained only by skilled monks with long meditation experience).

In addition to the standard frequency bands, Ray has found so called "focused arousal" frequency component at 38Hz. This frequency component was found during the Dharana stage of Rajayoga [26]. Ray supposed that it could represent possible functional component in the process of attention (Dharana means holding the mind at a certain point).

Changed perception during meditation is frequently reported. Subjects usually define it as a relaxed awareness with stable reception. We defined this state as dissociation of perception from the external sense organs. Quantitative investigation of this phenomenon is performed by Hirai, and alpha block dehabituation was found [22].

Particularly hard problem is analysis of transcendent signal. Ray defined it as a signal that is not bound to the time frame by any law of time domain [26]. He investigated transcendent signal in relation to the highly amused states of a child as well as state of deep aesthetic appreciation [27]. However, the transcendency is likely to be correlated with the clock of the organic system. These states are characterized by large number of impulses (spikes), and increased low frequency waves (theta and specially delta waves).

Fast wave activity was occasionally reported [24,25,27]. Banquet identified synchronous beta waves from all brain regions of almost constant frequency and amplitude [24]. That activity was found at four advanced meditants during the subjectively reported deepest meditation. Das and Gastaut performed electroencephalographic examination of seven yogis and observed that as the meditation progressed the alpha waves gave way to fast-wave activity at the rate of 40-45Hz, and that these waves subsided with a return of the slow alpha and theta waves [25]. Ray has found unusually large activity in the frequency range 16-18Hz, during highly amused states as well as state of deep aesthetic appreciation [27].

To the best of our knowledge EEG changes related to the healing process are rarely investigated. Zhang reported the EEG alpha activity during the Qi Gong state that occurred predominantly in the anterior regions. The peak frequency of EEG alpha rhythm was slower than the resting state, and the change of EEG during Qi Gong between anterior and posterior half had negative correlation. It can be seen that reported changes are very similar to the previously described changes during the meditation.

Proposed Methodology

We propose a methodology of characterizing certain neurophysiological states by tracing characteristic spatio-temporal patterns of EEG activities, as depicted in Figure 2 .

Figure 2 Block diagram of adopted methodology

The analysis is divided in two parts: static and dynamic.

  1. Static analysis uses artifact-free EEG to characterize long-term (average) activity. However, by removing signal sections with artifacts we lose temporal information as well.
  2. Dynamic analysis is performed on original signal to trace temporal patterns of activities as well as short-term changes in brain activities.

The analysis starts by expert's off-line manual artifact removing. In spite of some promising results in automatic artifact removal, manual removing using expert's knowledge is still preferred method in analysis. Then, topographic maps of artifact-free signal are built to indicate channels that have dominant activity in certain frequency bands (delta, theta, alpha, beta, etc.). The most interesting EEG channels are used for further signal processing procedures (spectral, coherence, wavelet, chaos and other analyses).

Then, on selected channels we perform dynamic analysis by constructing graphs with temporal dependencies of selected parameters (spectrogram, dominant band frequency, animation of topographic maps, coherence, ...).

On the other hand, dynamic analysis can indicate time intervals with significant changes of basic parameters (mean frequency, intensity, etc.), which are then subjected to additional static analysis.

According to our experience this interaction between static and dynamic analysis yields the best characterization of underlying neurophysiological changes.

In addition, interdependence analysis provides subtle information on simultaneous changes in brain electrical activity recorded from two subjects in the interactive state of mind.

Although the frequency domain analysis represents indispensable signal analysis procedure, we have found very useful time-domain analysis on different frequency band limited signals. It emphasizes both short-time signal changes, as well as statistical properties of the signal.

Certain frequency bands may indicate activity on different hierarchical levels, as depicted in Table 1. Source of activity in gamma, beta and alpha frequency band is thoroughly investigated, and we introduce hypothetical framework of analysis for the activity in theta, delta and sub-delta bands. Our hypothesis follows direction of the higher three bands that lower frequency represents higher level of integration, i.e. information binding. Therefore, activation in certain frequency band may indicate activity on the equivalent consciousness level. The proposed scheme may correspond with Jung's structure "ego-consciousness-individual unconsciousness-collective unconsciousness".

Frequency band

Activity

 

Gamma

Individual neurons
 

Beta

Specialized regions
 

Alpha

Physical consciousness

Theta ( q )

Mental consciousness

Delta ( d )

Higher level of consciousness

sub - Delta

Collective consciousness

Table 1: Possible sources of activity in certain EEG frequency bands

Basic characteristics of the environment

In spite of extensive support for standard EEG signal processing, existing software packages do not provide enough flexibility for studying subtle EEG changes. Therefore, we decided to develop our own open software environment for at least two reasons: a) total control of procedure parameters, and b) the possibility to develop original and improve existing signal processing algorithms.

Our open software environment is called STATE (Spatio-Temporal EEG Alteration Tracing Environment) [30-31]. Although it was primary designed to provide support for signal processing functions, the great deal of efforts was spent to make efficient visualization procedures and methods. The realized software environment was developed to support proposed methodology of tracing subtle EEG changes.

The STATE software package is an interactive open environment, developed under Windows 3.11. Most procedures are executed using MATLAB 4.0 [33], and some procedures are developed in C language and integrated in the environment. Procedures provide the following support (for more details see [30-32]):

  • Spectral and correlation analysis of EEG (with optional removing of artifacts). For epoch length and introduced considerations see [34,35]

  • Spectrogram analysis

  • Cepstrum analysis

  • Topographic mapping (for details see [35]) of various parameters such as:

    • absolute and relative power in frequency bands

    • power ratios between bands

    • z score values

    • coherence

  • Monitoring of temporal changes of relevant spatial characteristics (cross-correlation values, animation of topographic maps, instantaneous envelope and frequency, amplitude and frequency modulation index, ...)

  • Wavelet analysis (decomposition on the wavelet orthonormal basis using different types of filters)

  • Chaos analysis (the correlation dimension of strange attractors)

Time domain analysis makes use of different signal processing techniques for the extraction of instantaneous envelope and phase of EEG signal [37]. The instantaneous envelope is proportional to the instantaneous root mean square value of the signal, and therefore the energy of a certain frequency band could be traced in time.

The EEG signal could be analysed as both amplitude and frequency modulated [37]. This type of analysis was used to quantify alpha modulation in relation to cerebral blood flow [38], but we have found it very useful to quantify EEG changes in both subjects during the healing session.

The most important signal parameters for characterizing altered states of consciousness in time-domain analysis are:

  • Histogram of temporal channel activation (based on current envelope value)

  • Signal envelope periodicity

  • Instantaneous frequency of certain frequency band

  • Band power peak to peak interval [39]

  • Amplitude modulation index

  • Frequency modulation index

Experimental set-up

We performed recordings in electromagnetically shielded room (Faraday cage) on 18 channel EEG machine MEDELEC 1A97, and obtained 16 channel EEG with common (average) reference. Electrode positions were adopted according to International 10-20 System: F7, F8, T3, T4, T5, T6, Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2. We used Ag/AgCl electrodes, with impedance less than 5KW . Bandpass filter was set to 0.5-30Hz, and power supply notch filter was not used. For AD conversion we used PC AT with ADC board Data Translation 2801 (16 channel) with the sampling frequency of 128Hz. Software for standard topographic (off-line) EEG analysis was RHYTHM V.8.0 from Stellate Systems Inc., Quebec, Canada. For spectral analysis we used 20-seconds segments of EEG data with optional off-line manual artifact removal.

EEG was recorded separately from two adult human subjects (healer and patient-healee), before, during and after the healing session for 120 seconds in each period. Patient was in relaxed state with eyes closed. Healer kept eyes closed and had no activity apart from mental effort. Subjects had no physical contact.

During the healing session healer had stable basic physiological parameters: heartbeat rate (72 beats per minute), breath (4 per minute) and almost ceased eye movements.

Results of analysis

The analysis on brain electrical activity changes led to the following conclusions:

  1. Healer's brain electrical activity showed shift in power spectrum toward slow frequencies (most pronounced in delta and theta range as illustrated in Fig. 3) during the session, compared to periods pre and after it. Spatial distribution of changes was dominant over frontal and anterior temporal regions, and symmetrical. This change is not usual correlate (in absence of drowsiness) of intense mental effort [3].

    2. Figure 3: Topographic maps of 20 seconds of artifact-free healers EEG in theta frequency band before (left) and during the healing session (right).

  2. Dcrease of alpha activity, and increase of maximum alpha frequency during and immediately after the healing session (see Figures 4 and 5). Both characteristics differ significantly from the meditation-based state of consciousness [22-25].

    3.  

    Figure 4: Spectral power of healer's channel F3 before and during the healing session (dashed line); Artifact free sections; Session time (80-100s)

    Figure 5: Spectral power of healer's channel F3 before and immediately after the healing session (dashed line); Artifact free sections;

  3. Reduced number of discrete frequency components, representing stabilization of electrical activity in frequency domain.

  4. Coherence of BEA from homologue left and right regions (8 pairs of electrodes), as a measure of functional coupling, showed profound changes in its pattern. Coherence of prefrontal and frontolateral region's BEA [3] (F7/F8, Fp1/Fp2) was significantly increased in delta and theta range during the session, and return to pre-session values only in F7/F8 channels. Frontal parasagital region (F3/F4) conversely showed the coherence increases only after the session and only in theta range. These patterns of cortical functional organization that occurred during defined time intervals clearly differ from one another, indicating characteristic brain regional activation/deactivation during specific tasks.

  5. The pattern of spatial activity found on healer's EEG was similarly induced in patient's EEG during the session.

  6. Subject's (healee) report on behavioral changes during the session indicated that there were rhythmical changes in his jaw muscle tone, with approximate frequency bellow 1Hz. EEG changes that could correspond to this phenomenon cannot be analyzed by conventional methods, but our envelope analysis of some frequency bands indicates possible confirmations for such changes. We suggest further exploration of these low-frequency phenomena.

  7. Significant stabilization of instant frequency in lower frequency bands, as depicted in Figure 6.

    8.  

    Figure 6: Instantaneous frequency of bandpassed filtered delta (1.5-2Hz) before, during and after the healing session;Artifact free sections; Session time (80-100s); Channel F3.

  8. Stabilization of energy fluctuation at lower frequencies, according to envelope changes (see Figure 7). Temporal envelope changes during the healing session and after it are shown. Increase of mean power in delta band should be observed.

    9. Figure 7: Envelope of bandpassed filtered delta (1.5-2Hz) before, during and after the healing session;Artifact free sections; Session time (80-100s); Channel F3.

  9. Occasional patterns of synchronous change in signal modulation as presented in Figure 8.

  10. Stable modulation of delta frequency band with approximately five seconds period.

Figure 8: Synchronous spatial change of bandpassed filtered delta envelope (1.5-2Hz) during the healing session;Channels F3 and T4; Artifact free sections; Session time (80-100s).

 

Conclusion

Studying states of consciousness require carefully chosen methodology and subtle processing and visualization procedures. Robust long term analysis inherently omits short-period analysis in brain activity. Moreover, care should be taken to avoid disclosing significant changes as artifacts. Therefore we suggest carefully chosen methodology as a combination of static and dynamic analysis, and "marking" instead of disclosing artifact-like changes. The proposed methodology and developed software environment (STATE) were used to analyze neuropsychological changes during healing session.

We have found that EEG changes during healing session provide significant base for the analysis of altered state of consciousness, as well as non-sensory interactions. Naturally, it is not easy to find right subjects for the experiments, but we have found few subjects that exhibit statistically significant changes.

The analysis emphasized significance of tracing spatio-temporal EEG changes, and particularly temporal tracing of signal modulation parameters. It was shown that this approach point out some very low frequency changes (bellow 1 Hz), that would be otherwise missed using standard computerized EEG analysis.

Further investigation will be directed toward selecting the most significant statistical parameters within the larger set of experiments, and precise quantification of interdependence correlates.

References

[1] A. Gevins, A. R幦ond, (Eds.), Handbook of EEG and Clinical Neurophysilogy, Vol.1: Methods of analysis of brain electrical and magnetic signals, Elsevier Science Publishers, Amsterdam, Holland, 1987.

[2] F.H. Lopes da Silva, W.S. van Leeuwen, A. R幦ond, (Eds.), Handbook of EEG and Clinical Neurophysilogy, Vol.2: Clinical applications of computer analysis of EEG and other neurophysiological signals, Elsevier Science Publishers, Amsterdam, Holland, 1986.

[3] E. Basar, EEG Brain Dynamics, Elsevier, Amsterdam, 1980.

[4] E. Basar, (ed.), Dynamics of Sensory and Cognitive Processing by the Brain, Springer Verlag, Berlin, Germany, 1988.

[5] Crick, F., The astonishing hypothesis. The scientific search for soul, Charles Scribner's Sons, and Maxwell Macmillan International, 1994.

[6] C.M. Gray, P. Kéoµig, A.K. Engel, W. Singer, "Oscillatory responses in visual cortex exhibit inter-columnal synchronization which reflects global stimulus properties", Nature 338, pp. 334-337, 1989.

[7] E.R. John, T. Yang, A.B. Brill, R. Young, K. Ono, "Double-labeled metabolic maps of memory", Science 233, pp. 1167-1175, 1986.

[8] G.C. John, "Switchboard versus statistical theories of learning and memory", Science 177, pp. 850-864, 1972.

[9] M.P. Stryker, "Is grandmother an oscillation?", Nature 338, pp. 297-298, 1989.

[10] W. Singer, "Synchronization of cortical activity and its putative role in information processing and learning" Annu. Rev. Physiol., 55, pp. 349-374, 1993.

[11] S. H. Strogatz, I. Stewart, "Coupled Oscillators and Biological Synchronization", Scientific American, December 1993, pp. 68-75, 1993.

[12] Tart, C., ed. Altered states of consciousness, Academic, New York, 1972.

[13] Satyanarayana, M., Rajeswari, K.R., Rani, N.J., Krishna, C.S., Rao, P.V., "Effect of Santhi Kriya on certain psychophysiological parameters: a preliminary study", Indian J. Physiol. Pharmacol., April 1992, 36(2), pp. 88-92, 1992.

[14] John, R.G., "The persistent paradox of psychic phenomena: an engineering perspective", Proc. IEEE, pp 136-170, 1982.

[15] Adey, W.R., "Tissue interactions with nonionizing electromagnetic fields", Physiol. Rev., 61, pp. 435-514, 1981.

[16] D. Rakovi, "Neural networks, brainwaves, and ionic structures: Acupuncture vs. altered states of consciousness", Acunp. & Electro-Therap. Res., Intl. J., 16, pp. 89-99, 1991.

[17] D. Rakovi, "Neural networks vs. brainwaves: Biophysical model for ELF interactions", Proc. Ann. Int. Conf. IEEE/EMBS 14, pp. 2750-2751, 1992.

[18] D. Rakovi, "Brainwaves, neural networks, and ionic structures: Biophysical model for states of consciousness", this book.

[19] Rakovi, D., Fundamentals of Biophysics, Grosknjiga, Beograd, 1994.

[20] K.C. Markides, Fire in the Heart. Healers, Sages and Mystics, Paragon House, New York, U.S.A., 1990.

[21] Y. Omura, T.L. Lin, L. Debreceni, B.M. Losco, S. Freed, T. Muteki, C.H. Lin, "Unique changes found on the qi gong (chi gong) master's and patient's body during qi gong treatment: Their relationship to certain meridians & acupuncture points and the re-creation of therapeutic qi gong states by children & adults, Acunp. & Electro-Therap. Res. Int. J., 14, pp. 61-89, 1989.

[22] T. Hirai, "Electroencephalograpic Study of Zen Meditation", Psychiat. Neurol. Jap., 62, pp. 76-105, 1960.

[23] Banquet, J.P., "EEG and Meditation", Electroencephalography and Clinical Neurophysiology, Vol. 33, pp. 449-458, 1972.

[24] Banquet, J.P., "Spectral Analysis of the EEG in Meditation", Electroencephalography and Clinical Neurophysiology, Vol. 35, pp. 143-151, 1973.

[25] R.K. Wallace, H. Benson, "The Physiology of Meditation", Scientific American, 226:84, 1972.

[26] G.C. Ray, "Higher Stages of Rajayoga and Its Possible Correlation with Process of Evolution", Journal of the Institution of Engineers (India), ID 68, pp. 37-42, 1988.

[27] G.C. Ray, A.Y. Kaplan, "Transcendent Signal and its Possible Signature on Electroencephalogram", Journal of the Institution of Engineers (India), Vol. 74, March 1994, pp. 22-31, 1994.

[28] G.C. Ray, "r-Parameter: A New Measure of Level of Consciousness", unpublished paper.

[29] Zhang, J.Z, Zhao, J., He, Q.N., "EEG findings during special psychical state (Qi Gong State) by means of comressed spectral array and topographic mapping", Computers in Biology and Medicine, Vol. 18, No. 6, pp. 455-463, 1988.

[30] E. Jovanov, D. Rakovi, V. Radivojevi}, D. Ku{i}, P. [ukovi}, M. Car, "Evaluation of state of consciousness using software support for monitoring spatio-temporal EEG changes", Proc. ISCA Int. Conf. on Computer Applications in Engineering and Medicine, 1995.

[31] D. K, Integration of the software environment for EEG analysis, B. Sc. thesis, Faculty of Electrical Engineering, University of Belgrade, 1994.

[32] P. ukovi, EEG correlates of the altered states of consciousness, B. Sc. thesis, Faculty of Electrical Engineering, University of Belgrade, 1994.

[33] MATLAB 4.2 User's guide, The MathWorks, Inc., Natick, U.S.A., 1994.

[34] Jervis, B.W., Coelho, M., Morgan, G.W., "Spectral analysis of EEG responses", Medical&Biological Engineering&Computing, May 1989, pp. 230-238, 1989.

[35] Cohen, B.A., Sances, A., "Stationarity of the human electroencephalogram", Medical&Biological Engineering&Computing, September 1977, pp. 513-518, 1977.

[36] F.H. Duffy, (ed.), Topographic Mapping of Brain Electrical Activity, Butterworths, Boston, U.S.A., 1986.

[37] Ktonas, P.Y., Papp, N., "Instantaneous envelope and phase extraction from real signals: Theory, implementation and an application to EEG analysis", Signal Processing, Vol. 2, pp. 373-385, 1980.

[38] Okyere, J.G., Ktonas, P.Y., Meyer, J.S., "Quantification of the Alpha EEG Modulation and Its Relation to Cerebral Blood Flow", IEEE Transactions on Biomedical Engineering, Vol. BME-33, No. 7, pp. 690-696, 1986.

[39] E. Jovanov, D. Rakovi, V. Radivojevi}, D. Ku{i,"Band power envelope analysis - a new method in quantitative EEG", submitted to IEEE EMBS Conference

21st, The VXM Network, https://vxm.com

s