CN105188525A - Method and system to calculate quantitative EEG - Google Patents

Abstract

A system (20) and method (600) for calculating a quantitative EEG is disclosed herein. EEG signals are generated from an EEG machine (40) comprising electrodes (35), an amplifier (42) and processor (41). The EEG signals are processed continuously for artifact reduction to generate continuous artifact reduced EEG data. A quantitative EEG is computed using continuous artifact reduced EEG data in near real time.

Description

For the method and system of calculation in quantity EEG

Technical field

Present invention relates in general to the method and system for calculation in quantity EEG.

Background technology

Electroencephalogram (" EEG ") is that the electrical activity of the brain of measurement and recorder is to evaluate the diagnostic tool of brain function.Multiple electrode is attached to the head of people and is connected to machine by wiring.Machine amplifying signal the electrical activity of the brain of recorder.By producing electrical activity to the neural activity summation on multiple neuron.These neurons generate little voltage field.The gathering of these voltage fields produces the electric reading that the electrode on the head of people can detect and record.EEG is the superposition of multiple better simply signal.In adult normal, the amplitude of EEG signal is usually in the scope of 1 microvolt to 100 microvolts, and EEG signal is about 10 millivolts to 20 millivolts when utilizing cerebral dura mater bottom electrode to measure.The information relevant with the medical conditions of potential neural activity and people is provided to the amplitude of the signal of telecommunication and the monitoring of Time dynamic.

EEG be performed with: diagnosis epilepsy; Verify the problem about loss of consciousness or dementia; The cerebration of the people in checking stupor; Research sleep disorder, movable at surgery monitor cerebral, and extra physical problems.

Multiple electrode (usual 17 to 21, but exist for the normal place of at least 71) is attached to the head of people during EEG.Electrode is quoted by the position of the relevant electrode of the cerebral lobe of the brain with people or region.Quote as follows: F=forehead; Fp=antinion; T=temporal lobe; C=central authorities; P=calvarium; O=occipitalia; And (ear's electrode) of the nearly ear of A=.Numeral is used to make position narrow further, and " z " puts the electrode site related in the center line of the head of people.Electrocardiogram (" EKG ") can also be apparent on EEG display.

EEG uses the various combinations being referred to as the electrode of lead (montages) to record brain wave from different amplifier.The clear picture of the spatial distribution being generally created the EEG be provided on cortex of leading.Lead is the electrograph and the particular combination preferably referred at the checked electrode of particular point in time that obtain from the space array of recording electrode.

In bipolar lead, link continuous print electrode pair by the input 1 electrode input 2 of a passage being connected to passage subsequently, make adjacency channel have a common electrode.The bipolar chain of electrode can from front to back (longitudinal) or from left to right (horizontal) connect.In bipolar lead, compare the signal between two active electrode sites, obtain the difference of the activity of recording.Leading of another kind of type is that reference is led or single-stage is led.With reference in leading, various Electrode connection is to the input 1 of each amplifier and reference electrode is connected to the input 2 of each amplifier.In reference is led, collect signal in active electrode site and this signal and common reference electrode are compared.

With reference to leading the true amplitude and form that are of value to and determine waveform.For temporal lobe electrode, the scalp reference that CZ is normally good.

It is crucial for can positioning (" location ") to the source point of electrical activity for analyzing EEG.Usually come by identification " phasing back " the location of the normal or abnormal brain electric wave in bipolar lead, the deflection of two passages that " phasing back " refers in chain refers in the opposite direction.In reference is led, all passages can illustrate deflection in the same direction.If the electrical activity at active electrode place when be positive when the Comparison of Gardening Activities at reference electrode place, then deflection will downwards.The electrode that wherein electrical activity is identical with at the electrical activity at reference electrode place will not illustrate any deflection.Usually, there is the maximum electrode upward deflected represent with reference to the maximum negative activity in leading.

The instruction of some patterns is towards the tendency of the epilepsy in people.These ripples can be called by doctor " epileptic abnormal " or " epilepsy wave ".These comprise spike (spike), sharp wave (sharpwave) and spike electric discharge (spike-and-wavedischarge).Spike in the specific region of the such as brain of left temporal lobe and sharp wave indicating section epilepsy may likely come from this region.On the other hand, PGE is discharged by the spike be extensively distributed on two hemisphere of brain and discloses, especially when spike electric discharge starts in two hemisphere simultaneously.

There is the brain wave of some types: alpha ripple, beta ripple, delta ripple, theta ripple and gamma ripple.Alpha ripple has the frequency of 8 hertz to 12 hertz (" Hz ").Alpha ripple is usually closed when still people is vigilance in spirit when the eyes of people when people loosens or in waking state and is found.When the eyes of people open or people focuses one's attention on, Alpha ripple terminates.Beta ripple has the frequency of 13Hz to 30Hz.Beta ripple finds when people's vigilance, thinking, excitement or when taking the certain drug of high dose usually.Delta ripple has the frequency being less than 3Hz.Delta ripple usually when people fall asleep (non REM sleep or do not have have a dream sleep) or people is child time find.Theta ripple has the frequency of 4Hz to 7Hz.Theta ripple finds when people falls asleep (having a dream or REM sleep) or people is child usually.Gamma ripple has the frequency of 30Hz to 100Hz.Gamma ripple finds when higher mental activity and motor function usually.

Use herein to give a definition.

The vertical dimension that " amplitude " refers to from trough to maximum peak (negative or positive) records.It represents the information relevant with its activation synchronicity between component generation with the size of neuron pool.

Term " Analog-digital Converter " refers to the situation when analogue signal is converted into the digital signal that can be stored afterwards in a computer for process further.Analogue signal is " real world " signal (such as, the physiological signal of such as electroencephalogram, electrocardiogram or electro-oculogram).Be stored to make them and handled by computer, these signals must be converted into the discrete digital form that computer can be understood.

" artifact (artifact) " is the signal of telecommunication still stemming from non-brain source point detected along scalp by EEG.There is patient's related artifacts (such as, movement, perspiration, ECG, eye move) and technology artifact (artifact that 50/60Hz artifact, cable move, electrode cream is relevant).

Term " difference amplifier " refers to the key of electro physiology instrument.It is amplified in the difference (often pair of electrode amplifier) between two inputs.

" persistent period " is the interval turning back to baseline from change in voltage to it.It is also the measurement result that the neuronic synchronicity related in component generates activates.

" electrode " refers to for setting up the conductor with the electrical contact of the non-metallic part of circuit.EEG electrode is the lamellule be usually made up of rustless steel, stannum, gold or the silver that is coated with silver oxide coating.They are placed in the special position on scalp.

" electrode gel ", as the plastic extension of electrode, makes the movement of contact conductor unlikely produce artifact.Gel makes contact skin maximize and allow percutaneous Low ESR record.

Term " positioning of electrode " (10/20 system) refers to the standardized arrangement of the scalp electrode recorded for typical EEG.The essence of this system is the distance in the percent of 10/20 scope between the nasion-occipital protuberance (Nasion-Inion) and fixing point.These points are marked as antinion (Fp), central authorities (C), calvarium (P), occipitalia (O) and temporal lobe (T).Central electrode has been labeled footnote z, and it represents zero.Odd number is used as the footnote of the point on left hemisphere, and even number is used as the footnote of the point on right hemisphere.

" electroencephalogram " or " EEG " refers to by the track of record from the brain wave made by electroencephalograph of the electrical activity of the brain of scalp.

" electroencephalograph " refers to for detecting and recording the device (being also called electroencephalography (encephalograph)) of brain wave.

" epileptic " refers to similar epilepsy.

" filtering " refers to the process removing undesired frequency from signal.

" wave filter " is the equipment of the frequency component of change signal.

" lead " and refer to the layout of electrode.Bipolar lead or reference can be utilized to lead to monitor EEG.Bipolar lead refers to that each passage exists two electrodes, therefore there is reference electrode for each passage.Refer to that collective reference electrode is existed for all passages with reference to leading.

" form " refers to the shape of waveform.The shape of ripple or EEG pattern are by combining the frequency that forms waveform and being determined by their phase place and voltage relationship.Ripple pattern can be described to: significantly EEG is movable, that comprise the multiple frequencies being combined to form complicated wave form " polymorphic " remarkable EEG is movable, " sinusoidal " ripple of similar sine wave to be revealed as " monomorphism " comprising a dominant activity.Monomorphism activity is visibly different sinusoidal " transition " solitary wave or pattern with background activity normally.

" spike " refers to have spike and from 20 milliseconds to the transition lower than persistent period of 70 milliseconds.

Term " sharp wave " refers to have the transition of the persistent period of spike and 70 to 200 milliseconds.

Term " neural network algorithm " refers to the algorithm identifying and have the sharp wave transition of the high likelihood being epileptic exception.

" noise " refers to any undesired signal of the signal that amendment is expected.It can have multiple source.

" periodically " pattern or element distribution in time (such as, the movable appearance on more or less regular interval of specific EEG) is referred to.Activity can be general, focal or inclined side property.

EEG epoch (epoch) is the amplitude of the EEG signal according to time and frequency.

Once certain time used quantitative EEG (QEEG) in the analysis of EEG in the past.The images outputting of the Time Compression that the most frequently used is for using FFT.Such images outputting can understand to illustrate the such as long-time general survey of section EEG in frequency range by human reader.Although single page EEG can show the data of ten seconds, one page QEEG can display minute or even a few hours.

QEEG can also for generation of with the time averaging result of single digit value, at some preset time.This can be equally simple with average amplitude.Or it can be the result of calculation of the ripple be limited in single frequency scope.

QEEG can be limited to the subset of the quantity of the passage of record.By this way, calculate the activity in reflection hemisphere, or the smaller portions of brain.

Described result of calculation also can be calculated as the relative value of two subsets of passage or two different frequency ranges.Thought is that the change of these relative value can be important in diagnosis.

Exist use QEEG to understand a large amount of academic interests of EEG.Design is that QEEG is not more subjective and faster than checking potential waveform.Pattern also may appear in one's mind in time, and this pattern is also difficult to see if not seeing.

Example is the diagnosis to apoplexy.Should believe, when apoplexy starts, the change of cerebration is almost reflected in EEG immediately.This will appear at significantly in many cases before there is clinical symptoms.Therefore, exist the continuous monitoring of the patient being in stroke risk to provide the great interest of early diagnosis and therapy.

But, be huge to the obstacle of monitoring continuously.The first, monitoring raw EEG signal is very labour-intensive continuously.The second, the type of little relative change of reflection apoplexy is very difficult to observe, especially when once with the only data of ten seconds in current.QEEG can be to this solution and exist and attempt determining which kind of result of calculation can illustrate the research of the important well afoot of the type of the change of reflection apoplexy.But work is in this field defeated very greatly because very a large amount of existence of the artifact in EEG has been subjected to.

In Scalp EEG signals, such as muscle, eye move and can flood the signal for brain by the artifact of the bad electrical contact of electrode.Expert's level reviewer is known and is ignored these artifacts and pay close attention to without artifact sections, but QEEG does not have this luxury and all signals are in the calculation involved.Result is that QEEG usually reflects that artifact as many or than it reflects that cerebration is more with its reflection cerebration.Certainly, this is problematic when producing graphic result, but expert's level reviewer may can pick out the pattern deriving from cerebration again in that case.But when calculating centrifugal pump in order to diagnostic purpose, this is very large problem.For this reason, researcher is attempted selecting relatively artifact-free fragment continually and is calculated, but this is not useable in clinical practice certainly.

Therefore, especially in clinical setting, there is the demand to the QEEG containing complete signal but the artifact greatly reduced.

Summary of the invention

This solution is the many artifacts removed with account form before QEEG process in the artifact existed in record.By this way, significantly can improve signal to noise ratio, and the QEEG obtained calculates and will react cerebration.In this, can determine that the QEEG of which kind of type will be effective and use it clinically in diagnosis afterwards simultaneously.

There is the research estimating (QEEG) to expect the clinical symptoms of apoplexy calculated and discussion that may use EEG in the art.

QEEG is used to be that artifact produces insecure quantitative values when being mixed into brain signal in the subject matter expecting the clinical symptoms of apoplexy about carrying out.The present invention achieves the level that the artifact that QEEG is practical is now reduced on the basis of monitoring continuously.

Exemplarily, in apoplexy diagnosis, doctor can start being confirmed as being the continuous monitoring of the one or more QEEg parameters having diagnostic value.Establish baseline, and if the doctor scope QEEG that can set for these parameters moves to staff outside these scopes will be warned possible apoplexy.In the embodiment of more automatization, system can determine baseline and automatically set point, or it can use the intelligence system of such as neutral net to determine the QEEG that will use and the one group of change representing apoplexy.Apoplexy is only single example, and can diagnose other situations many affecting cerebration by this way.

Accompanying drawing explanation

Fig. 1 is the image of quantitative EEG.

Fig. 2 is the figure of the system for calculation in quantity EEG.

Fig. 3 is the figure for the arrangement of electrodes for EEG.

Fig. 4 is the details drawing for the arrangement of electrodes for EEG.

Fig. 5 is that CZ is with reference to the diagram of leading.

Fig. 6 is the diagram of the EEG record containing epilepsy, muscle artifact and eye motion artifact.

Fig. 7 is the diagram of the EEG record of wherein Fig. 6 of being removed of muscle artifact.

Fig. 8 is the diagram of the EEG record of wherein Fig. 7 of being removed of eye motion artifact.

Fig. 9 is the flow chart of the method for calculation in quantity EEG.

Figure 10 is the flow chart of the method for calculation in quantity EEG.

Figure 11 is the figure of the system for calculation in quantity EEG.

Detailed description of the invention

Figure 1 illustrates the image 100 of quantitative EEG (" qEEG ").Described method and system allows the EEG record according to reducing through artifact to generate qEEG, and without the need to removing part that EEG records to prevent artifact effects qEEG.

Fig. 2 illustrates the system 20 for calculation in quantity EEG.Patient 15 has on electrode cap 31, electrode cap 31 comprises multiple electrode 35a-35c, multiple electrode 35a-35c are attached to the head of patient, wherein, wiring 38 is connected to EEG machine part 40, EEG machine part 40 from electrode 35 and comprises amplifier 42, and amplifier 42 is for being amplified to the signal of computer 41, computer 41 has processor, and described processor also generates the EEG record 51 and qEEG that can observe on display 50 for the signal analyzing self-electrode 35.The U.S. Patent number more detailed description of the electrode that the present invention utilizes being described in detail in the people such as Wilson is in " aMethodAndDeviceForQuickPressOnEEGElectrode " of 8112141, its entirety is incorporated to by reference herein.EEG is optimized to carry out automatic artifact filtration.Use neural network algorithm to be processed to generate treated EEG record after EEG record, described treated EEG records for generating qEEG.

The U.S. Patent Application No. in JIUYUE in 2012 submission on the 15th extra description analyzing EEG record being set forth in the people such as Wilson is in " aMethodAndSystemForAnalyzingAnEEGRecording " of 13/620855, its entirety is incorporated to by reference herein.

Patient has multiple electrodes of the head being attached to patient, and wherein wiring is from Electrode connection to amplifier, and described amplifier is for being amplified to the signal of processor, and described processor is for analyzing the signal of self-electrode and creating EEG record.The difference of brain on the head of patient produces unlike signal.Multiple electrode is positioned on the head of patient, as shown in Figure 3 and Figure 4.CZ site is in center.Such as, on Fig. 4, Fp1 is present in the passage FP1-F3 on Fig. 6.The quantity of electrode determines the quantity of the passage of EEG.The passage of larger amt produces the more detailed expression of the cerebration of patient.Preferably, each amplifier 42 of EEG machine part 40 is corresponding to two electrodes 35 of head being attached to patient 15.From the output of EEG machine part 40 be by two electrode detection to electrical activity in difference.The layout of each electrode is crucial for EEG report, because electrode pair is the closer to each other, then the difference of the brain wave recorded by EEG machine part 40 is less.

EEG is optimized for and carries out automatic artifact filtration.Use neural network algorithm to be processed to generate treated EEG record after EEG record, described treated EEG records analyzed for display.Between the Harvest time recorded EEG, processing engine performs the continuous analysis of EEG waveform and by the existence of the electrode artifact basis of passage being determined most of type.More as human reader, processing engine detects artifact by the multiple features analyzing EEG track.Preferred artifact detects independent of impedance inspection.Between Harvest time, the passage that this process monitoring arrives is to find electrode artifact.When artifact being detected, described artifact automatically removes from epilepsy testing process and optionally removes from trend display.This obtains the epilepsy accuracy in detection of much higher level and is easier to reading trend than former generation product.

Algorithm for removing artifact from EEG uses blind source separating (BSS) algorithm of such as CCA (canonical correlation analysis) and ICA (independent component analysis) that the signal of one group of passage is transformed into a group component ripple or " source " usually.

In one example, the algorithm of BSS-CCA is called as removing the effect of musculation from EEG.Record lead on use described algorithm often will not produce the result of optimization.In this case, wherein reference electrode is generally used to be such as adopt one in the crown electrode of the CZ of international 10-20 standard lead to be optimum.In the algorithm, first leading of record will be transformed into CZ with reference to leading before artifact removes.Signal designation on CZ its when not being best selection, then described algorithm will be loaded into the list of possible reference electrode to find out applicable reference electrode.

Can directly user select lead on perform BSS-CCA.But this has two problems.The first, this require be selected for observed by user each lead on carry out expensive artifact and remove process.The second, artifact removes and changes between leading in difference, and will be only optimum when the reference of the optimum reference of user's choice for use is led.Because observe needed for EEG often lead different with for removing leading of optimum for artifact, so this solution not being.

Fig. 5-Fig. 8 illustrates how to remove artifact to allow the more clear diagram for the authentic activity of the brain of reader from EEG signal.Fig. 6 is the diagram of the EEG record 4000 containing epilepsy, muscle artifact and eye motion artifact.Fig. 7 is the diagram of the EEG record 5000 of wherein Fig. 6 of being removed of muscle artifact.Fig. 8 is the diagram of the EEG record 6000 of wherein Fig. 7 of being removed of eye motion artifact.

The various trend of EEG record are generated by processing engine.The inhibition ratio of epilepsy probability trend, rhythmicity spectrogram left hemisphere trend, rhythmicity spectrogram right hemisphere trend, FFT spectrogram left hemisphere trend, FFT spectrogram right hemisphere trend, asymmetric relative spectrogram trend, asymmetric adiabatic index trend, aEEG trend and left hemisphere trend and right hemisphere trend.

Rhythmicity spectrogram allows the differentiation of the epilepsy of observing in single image.Rhythmicity spectrogram measures the amount of the rhythmicity existed on each frequency in EEG.

Epilepsy probability trend illustrates the epileptic activity probability calculated in time.Epilepsy probability trend illustrates the persistent period of the epilepsy detected, and discloses and may drop to epilepsy and detect cut-off (cutoff) below but still be the region observing interested record.Epilepsy probability trend, when showing together with other trend, provides the comprehensive observing to the quantitative change in EEG.

As shown in Figure 9, a kind of method for calculation in quantity EEG is denoted as 600 generally.At frame 601, generate EEG signal from the EEG machine comprising multiple electrode, amplifier and processor.At frame 602, process EEG signal continuously and be used for artifact minimizing to generate treated EEG record.At frame 601, record calculation in quantity EEG according to treated EEG.Preferably, fast Fourier transform signal processing is used to carry out calculation in quantity EEG.Reduce artifact type be select from the group comprising the following: nictation artifact, muscle artifact, tongue motion artifact, chew artifact and heart beating artifact.

As shown in Figure 10, a kind of method for calculation in quantity EEG is denoted as 700 generally.At frame 701, generate EEG signal from the EEG machine comprising electrode, amplifier and processor.At frame 702, process EEG signal continuously and be used for artifact minimizing to generate the EEG data reduced through continuous artifact.At frame 703, use the close calculation in quantity EEG in real time of the EEG data reduced through continuous artifact.Described method also comprises expects apoplexy based on described quantitative EEG.Described method alternatively comprises and utilizes described quantitative EEG to carry out epilepsy detection.

Figure 11 and Figure 12 illustrates the system for calculation in quantity EEG.Patient 15 has on electrode cap 31, electrode cap 31 comprises multiple electrode 35a-35c, multiple electrode 35a-35c are attached to the head of patient, wherein wiring 38 is connected to EEG machine part 40 from electrode 35, EEG machine part 40 comprises amplifier 42, amplifier 42 is for being amplified to the signal of computer 41, and computer 41 has processor, and described processor also generates the EEG record and qEEG51 that can observe on display 50 for the signal analyzing self-electrode 35.CPU41 comprises for the software program of neural network algorithm and the software program for qEEG engine.As shown in figure 12, artifact reduces the parts of engine, qEEG engine 47, microprocessor 44, memorizer 42, Memory Controller 43 and I/O48 and EEEG machine 40.EEG is optimized for automatic artifact and filters.Use neural network algorithm to be processed to generate treated EEG record after EEG record, described treated EEG records analyzed for display.

Claims (10)

1., for a method of calculation in quantity EEG, described method comprises:

Multiple EEG signal is generated from the machine comprising multiple electrode, amplifier and processor;

Process described multiple EEG signal continuously to reduce for artifact to generate treated EEG record; And

Calculation in quantity EEG is carried out according to treated described EEG record.

2. method according to claim 1, wherein uses fast Fourier transform signal processing to calculate described quantitative EEG.

3. method according to claim 1, the artifact type wherein reduced selects from the group comprising the following: nictation artifact, muscle artifact, tongue motion artifact, chew artifact and heart beating artifact.

4., for a system of calculation in quantity EEG, described system comprises:

Multiple electrode, described multiple electrode is for generating multiple EEG signal;

Processor, described processor is connected to described multiple electrode to generate EEG record according to described multiple EEG signal; And

Display, described display is connected to described processor for display EEG record;

Wherein said processor is configured to carry out calculation in quantity EEG according to treated described EEG record.

5. system according to claim 4, wherein said processor is configured to use multiple neural network algorithm to process described EEG signal to create treated described EEG record.

6. system according to claim 5, the artifact type wherein reduced selects from the group comprising the following: nictation artifact, muscle artifact, tongue motion artifact, chew artifact and heart beating artifact.

7., for a method of calculation in quantity EEG, described method comprises:

Multiple EEG signal is generated from the machine comprising multiple electrode, amplifier and processor;

Process described multiple EEG signal continuously and reduce the EEG data reduced to generate continuous print artifact for artifact; And

The close calculation in quantity EEG in real time of the EEG data using continuous print artifact to reduce.

8. method according to claim 7, also comprises and expects apoplexy based on described quantitative EEG.

9. method according to claim 7, wherein uses fast Fourier transform signal processing to calculate described quantitative EEG.

10. method according to claim 7, also comprises and utilizes described quantitative EEG to detect for epilepsy.