Digital Analysis Of EEG Brain Signal - Analysis Of EEG Brain Signal. Author(s): Dubey R , Pathak A ....

Click here to load reader

  • date post

  • Category


  • view

  • download


Embed Size (px)

Transcript of Digital Analysis Of EEG Brain Signal - Analysis Of EEG Brain Signal. Author(s): Dubey R , Pathak A ....

  • Article ID: WMC001193 2046-1690

    Digital Analysis Of EEG Brain SignalCorresponding Author:Mr. Rash Dubey,Asst. Prof., E & IE, APJ College of Engg., Sohna, Gurgaon, 121003 - India

    Submitting Author:Dr. Rash B Dubey,Professor, ECE Dept, Hindu College of Engg, Sonepat, 121003 - India

    Article ID: WMC001193

    Article Type: Research articles

    Submitted on:19-Nov-2010, 06:02:11 AM GMT Published on: 20-Nov-2010, 01:40:52 AM GMT

    Article URL:

    Subject Categories:BRAIN

    Keywords:ECG brain signal, Fourier transform, Independent component analysis.

    How to cite the article:Dubey R , Pathak A . Digital Analysis Of EEG Brain Signal . WebmedCentral BRAIN2010;1(11):WMC001193

    Source(s) of Funding:

    not applicable

    Competing Interests:


    WebmedCentral > Research articles Page 1 of 19

  • WMC001193 Downloaded from on 22-Dec-2011, 12:01:36 PM

    Digital Analysis Of EEG Brain SignalAuthor(s): Dubey R , Pathak A


    Electroencephalography is the neurophysiologicmeasurement of the electrical activity of the brainusing electrodes placed on the scalp. The resultingtraces are known as electroencephalogram (EEG) andthey represent an electrical signal from a large numberof neurons. The EEG is a brain non-invasiveprocedure frequently used for diagnostic purpose.Instead of electrical currents the voltage differencesbetween different parts of the brain are observed. TheEEG consists of a set of multi-channel signals. Thepattern of changes in signals reflects large-scale brainactivities. In addition the EEG also reflects activationof the head musculature, eye movements, interferencefrom nearby electric devices, and changingconductivity in the electrodes due to the movements ofthe subject or physicochemical reactions at theelectrode sites. In the proposed work advance signalprocessing technique like Fast Fourier Transform andIndependent Component Analysis are used to analyzethe various brain activities. Both the techniques areapplied to single trail multi-channel EEG data. TheFast Fourier transform is to determine the powercontent of the frequency band and estimate thefrequency component while independent componentanalysis performs the blind source of separation ofstatistically independent source and separates thecomponents for their periods of activation


    Electroencephalography is a medical imagingtechnique that reads scalp electrical activity generatedby brain structures. The electroencephalogram (EEG)is defined as electrical activity of an alternating typerecorded from the scalp surface after being picked upby metal electrodes and conductive media. The EEGmeasured directly from the cortical surface is calledelectrocortiogram while when using depth probes it iscalled electrogram. Thus electroencephalographicreading is a completely non-invasive procedure thatcan be applied repeatedly to patients, normal adults,and children with virtually no risk or limitation.When brain cells are activated, local current flows areproduced. EEG measures mostly the currents that flow

    during synaptic excitations of the dendrites of manypyramidal neurons in the cerebral cortex. Differencesof electrical potentials are caused by summedpostsynaptic graded potentials from pyramidal cellsthat create electrical dipoles between soma (body ofneuron) and apical dendrites. Brain electrical currentconsists mostly of Na+, K+, Ca++, and Cl- ions thatare pumped through channels in neuron membranesin the direction governed by membrane potential thedetailed microscopic picture is more sophisticated,including different types of synapses involving varietyof neurotransmitters. Only large populations of activeneurons can generate electrical activity recordable onthe head surface. Between electrode and neuronallayers current penetrates through skin, skull andseveral other layers. Weak electrical signals detectedby the scalp electrodes are massively amplified, andthen displayed on paper or stored to computermemory. Due to capability to reflect both the normaland abnormal electrical activity of the brain, EEG hasbeen found to be a very powerful tool in the field ofneurology and clinical neurophysiology. The electricalsignals generated by the brain represent not only thebrain function but also the status of the whole body.The electrical nature of the human nervous systemhas been recognized for more than a century. It is wellknown that the variation of the surface potentialdistribution on the scalp reflects functional activitiesemerging from the underlying brain .This surfacepotential variation can be recorded by affixing an arrayof electrodes to the scalp, and measuring the voltagebetween pairs of these electrodes, which are thenfiltered, amplified and recorded [1]. Evoked potentialsor event-related potentials (ERPs) are significantvoltage fluctuations resulting from evoked neuralactivity. Evoked potential is initiated by an external orinternal stimulus [2]. Mental operations, such as thoseinvolved in perception, selective attention, languageprocessing and memory, proceed over time ranges inthe order of tens of milliseconds. Whereas PET andMRI can localize regions of activation during a givenmental task, ERPs can help in defining the time courseof these activations amplitudes of ERP componentsare often much smaller than spontaneous EEGcomponents, so they are not to be recognized fromraw EEG trace. They are extracted from set of singlerecordings by digital averaging of epochs of EEGtime-locked to repeated occurrences of sensory,cognitive, or motor events. The spontaneousbackground EEG fluctuations, which are random

    WebmedCentral > Research articles Page 2 of 19

  • WMC001193 Downloaded from on 22-Dec-2011, 12:01:36 PM

    relatively to time point when the stimuli occurred, areaveraged out, leaving the event-related brainpotentials. These electrical signals reflect only thatactivity which is consistently associated with thestimulus processing in a time-locked way. The ERPthus reflects, with high temporal resolution, thepatterns of neuronal activity evoked by a stimulus [4].EEG waves classification is achieved using anaccurate and highly distinguishable technique. Themethod makes use of both the discrete wavelettransform as well as the discrete Fourier transform.Specially, wavelet transform is used as a classifier ofthe EEG frequencies. In addition, the filtered EEG dataare used as input to the wavelet transform offers aperfect success in the rejecting undesired frequenciesand permits the discrete wavelet transform levels todiscriminate the EEG waves only [3]. EEG signals areconsidered not to be deterministic and they have nospecial characteristics like ECG signals. In addition,when the Fourier transform is applied to successivesegments of an EEG signal, the obtained spectra areobserved to be time varying. This indicates that theEEG signal is also non-stationary. The spectralanalysis based on the Fourier transform classicalmethod assumes the signal to be stationary, andignores any time-varying spectral content of the signalwithin a window [3].EEGLAB, runs under the cross-platform MATLABenvironment for processing collections of single-trialand/or averaged EEG data of any number of channels.Available functions include EEG data, channel andevent information importing, data visualization,preprocessing, independent component analysis (ICA)and time/frequency decompositions including channeland component cross-coherence supported bybootstrap statistical methods based on data re-sampling. EEGLAB functions are organized into threelayers. Top-layer functions allow users to interact withthe data through the graphic interface without needingto use MATLAB syntax. Menu options allow users totune the behavior of EEGLAB to available memory.Middle-layer functions allow users to customize dataprocessing using command history and interactive popfunctions [5].Decomposition of the EEG signal using ICA is arecently developed and practical technique for EEGdata analysis. ICA method determines source signalsfrom their mixture. This analysis allows us tounderstand the sources of EEG signal. The simpleexample with the real EEG data is considered in orderto resolve the sources of the artifacts and the sourcesof useful signal. It is also emphasized the clinicalsignificance of each component and hence theimportance of ICA method in clinical practice [8-9].

    Autoregressive (AR) spectral estimation techniquesare known to provide better resolution than classicalperiodogram methods when short segments of dataare select for analysis. It has been observed that theenergy in the EEG data segment is concentrated notin the beginning but somewhere in between the initialand the final positions thus confirming fact that EEG isa mixed delay signal. This position where the energy isconcentrated has been obtained with the help of leastsquares wave shaping filter. It is also shown that theknowledge of the position where the energy in thesignal is concentrated can be used in making a betterspectral estimation of short segments of EEG data [10].Single-channel blind source separation (BSS)technique can be used to decompose a single-channelrecording of brain activity into its constituentcomponents. This technique is used to identify andisolate rhythmic components underlying the recordings.In practice it is feasible to use band-pass filtering of aknown fixed frequency band for monitoring [11].Decomposition of single-trial multi-channel EEGrecordings onto temporally independent and spatiallystationary source signals, as well as identification andpossible removal of artifacts EEG recordings a