JP2016514022A5 - - Google Patents

Description

qEEG calculation method and system

  The present invention generally relates to a method and system for calculating quantitative EEG.

  An electroencephalogram (“EEG”) is a diagnostic tool that measures and records the electrical activity of the human brain to assess cerebral function. A plurality of electrodes are mounted on a person's head and connected to the device by wiring. The device amplifies the signal and records the electrical activity of the human brain. Electrical activity is generated by the sum of neural activity across multiple neurons. These neurons generate a small electric field. The sum of these electric fields forms an electrical reading that can be detected and recorded by an electrode on the person's head. EEG is a superposition of a plurality of simpler signals. For normal adults, the amplitude of the EEG signal is typically 1 microvolt to 100 microvolts, and the EEG signal is about 10 to 20 millivolts when measured with a subdural electrode. Monitoring electrical signal amplitude and transient power provides information about the underlying human neural activity and medical symptoms.

  EEG diagnoses epilepsy, demonstrates problems with loss of consciousness or dementia, demonstrates brain activity in a coma person, studies sleep disorders, monitors brain activity under surgery and additional physical problems To be executed.

  A plurality of electrodes (usually 17-21, but with more than 70 reference positions) are worn on the person's head during EEG. The electrodes are given a reference by the position of the electrode relative to the earlobe or human brain region. The reference marks are as follows. F = frontal; Fp = frontal; T = temporal; C = middle; P = parietal; O = occipital; A = ear (ear electrode). The position is further narrowed using numbers, and “z” refers to the electrode position of the centerline of the person's head. An electrocardiogram (“EKG”) can also be displayed on the EEG display.

  EEG records brain waves from different amplifiers using various combinations of electrodes called montages. In general, montages are created to provide a clear picture of the spatial distribution of EEG across the cerebral cortex. A montage is an electrical map obtained from a spatial array of recording electrodes, and preferably refers to a particular combination of electrodes that are investigated at a particular point in time.

  In bipolar montage, adjacent channels have one electrode in common because successive pairs of electrodes are related by connecting electrode input 2 of one channel to input 1 of the next channel. The bipolar chain of electrodes can be connected from front to back (vertical direction) or from left to right (lateral direction). In a bipolar montage, the signal between two operating electrode positions is compared to produce a recorded activity difference. Another type of montage is a reference montage or a monopolar montage. In the reference montage, various electrodes are connected to the input 1 of each amplifier, and the reference electrode is connected to the input 2 of each amplifier. In the reference montage, a signal is collected at the active electrode position and compared to a common reference electrode.

  The reference montage is suitable for determining the true amplitude and shape of the waveform. For temporary electrodes, CZ is usually a good scalp reference.

  The ability to locate the source of electrical activity (“stereolocation”) is important to EEG's analytical capabilities. Normal or abnormal localization of the electroencephalogram in a bipolar montage is usually accomplished by identifying "phase reversal", i.e., the deviation of two channels in the chain pointing in opposite directions. In the reference montage, all channels can show deviations in the same direction. If the electrical activity at the working electrode is positive compared to the activity at the reference electrode, the excursion is downward. An electrode whose electrical activity is identical to the reference electrode shows no excursion. In general, the electrode with the largest upward deflection represents the greatest negative activity in the reference montage.

Some patterns show a tendency toward human seizures. Doctors may refer to these waves as “epileptic abnormalities” or “epileptic waves”. This includes spikes, sharp waves, slow spikes, and the like. Spikes and sharp waves in certain areas of the brain, such as the left temporal lobe, indicate that partial seizures may have originated from that area. Meanwhile, epilepsy are mainly generalization, especially spike waves as long as it starts at the same time in both hemispheres of the brain, suggested by spike waves that are spread widely throughout both hemispheres.

  There are several types of brain waves, such as alpha waves, beta waves, delta waves, theta waves, and gamma waves. The frequency of the alpha wave is 8-12 hertz (“Hz”). Alpha waves are usually found when a person is relaxed or in a wakeful state with closed eyes but mentally alert. Alpha waves stop when a person's eyes are open or a person is concentrated. The beta wave has a frequency of 13 Hz to 30 Hz. Beta waves are usually found when a person is alert, thinking, upset, or taking a large amount of a specific drug. The delta wave has a frequency of less than 3 Hz. Delta waves are usually found only when a person is asleep (non-rem or dreamless sleep) or is an infant. Theta waves have a frequency of 4 Hz to 7 Hz. Theta waves are usually found only when a person is asleep (dream or REM sleep) or is an infant. The gamma wave has a frequency of 30 Hz to 100 Hz. Gamma waves are usually found during advanced mental activity and motor function.

  The following definitions are used herein.

  “Amplitude” refers to the vertical distance measured from the valley to the maximum peak (negative or positive). The amplitude represents information about the size of the neuron population and the starting simultaneity during component generation.

  The term “analog-to-digital conversion” refers to the conversion of an analog signal into a digital signal that is stored in a computer for further processing. An analog signal is a “real world” signal (eg, a physiological signal such as an electroencephalogram, electrocardiogram, or electrooculogram). In order to be stored and manipulated in a computer, these signals must be converted into a discrete digital form that the computer can understand.

  “Artifacts” are electrical signals that are detected along the scalp by EEG, but are derived from non-cerebrum. Artifacts include patient-related artifacts (eg, movement, sweating, ECG, eye movement) and technical artifacts (50/60 Hz artifact, cable movement, electrode paste related).

  The term “operating amplifier” refers to the key to electrophysiological instruments. The amplifier expands the difference between the two inputs (one amplifier per electrode pair).

  “Period” refers to the time interval from the start of the voltage change to the return to the baseline. It is also a measurement of the neuronal synchronization start that is included in the component generation.

  “Electrode” refers to a conductor used to establish electrical contact with a non-metallic portion of a circuit. The EEG electrode is a small metal disk usually made of stainless steel, tin, gold or silver coated with silver chloride. The electrodes are placed at specific locations on the scalp.

  Since the “electrode gel” plays a role as a malleable extension portion of the electrode, even if the electrode lead moves, artifacts hardly occur. The gel maximizes skin contact and allows low resistance recording through the skin.

The term “electrode arrangement” (10/20 system) refers to a standard arrangement of conventional scalp electrodes for EEG recording. The essence of this system divides the range between the nasal root point-exterior occipital ridge point and the fixed point by 10/20%. These fixed points are denoted as frontal (Fp), middle (C), parietal (P), occipital (O), and temporal (T). The centerline electrode is displayed with a subscript z representing zero. Odd as subscript for points to the left hemisphere, even numbers are used as subscripts for points in the right hemisphere.

  “Electroencephalogram” or “EEG” refers to EEG tracking by recording electrical activity of the brain from the scalp, performed using an electroencephalograph.

  An “electroencephalograph” refers to a device that detects and records an electroencephalogram (also referred to as an electroencephalogram).

  “Epileptic-like” refers to being similar to epilepsy.

  “Filtering” refers to the process of removing unwanted frequencies from a signal.

  A “filter” is a device that changes the frequency component of a signal.

  “Montage” means the placement of electrodes. EEG can be monitored in either a bipolar montage or a reference montage. Bipolar means that there are two electrodes in one channel, so there is a reference electrode for each channel. Reference montage means that there is a reference electrode common to all channels.

  “Form” refers to the shape of the waveform. The shape of the wave or EEG pattern is determined by the frequency and phase and voltage relationships that combine to form the waveform. A wave pattern is an individual EEG activity that appears to consist of one dominant activity, “singular”, an individual EEG activity that consists of multiple frequencies that combine to form a complex waveform, “polymorphism”, It can be described as a “sine” resembling a sine wave (a monomorphic activity is usually a sine wave), a “transient” that is an independent wave or pattern that is distinctly different from the background activity.

  “Spine wave” refers to a transient signal with a distinct peak and a duration of 20-70 msec.

  The term “sharp wave” refers to a transient signal with a distinct peak and a period of 70-200 msec.

  The term “neural network algorithm” refers to an algorithm that identifies a steep transient signal with a high probability of being an epilepsy-like abnormality.

  “Noise” refers to an unwanted signal that alters the desired signal. Noise can have multiple sources.

  “Regularity” refers to a time distribution of patterns or elements (eg, the appearance of specific EEG activity at somewhat regular intervals). Activities can be generalized, localized, or lateralized.

  The EEG epoch is the amplitude of the EG signal that is a function of time and frequency.

  Quantitative EEG (QEEG) has been used for the analysis of EEG for some time. The most common use is time-compressed graphical output that utilizes FFT. This type of graphical output can be interpreted by a human reader and can provide an overview of long-term EEG in the frequency range, for example. One page of EEG may display 10 seconds of data, while a page of QEEG may display several minutes or hours of data.

  QEEEG can also be used to produce time averaged results with a single number at a given time. This can be as simple as the average amplitude. Alternatively, the calculation can be limited to waves in a single frequency range.

QEEG can be limited to a subset of the number of recorded channels. Thus, the calculation reflects the activity of the hemisphere or a small part of the brain.

  The calculation may also be calculated as a relative value of two subsets of channels or two different frequency ranges. According to this concept, changes in these relative values are important for diagnosis.

  There is great academic interest in the interpretation of EEG using QEEEG. According to this concept, the subjectivity is lower and quicker than viewing the basic waveform. In addition, there are cases where a pattern that is impossible or difficult to visually observe by other methods may occur.

  One example is a diagnosis of stroke. At the start of a stroke, changes in brain activity are likely to be reflected in the EEG almost immediately. In many cases, this change occurs clearly before clinical symptoms appear. Accordingly, there is great interest in continuously monitoring stroke risk patients and providing early diagnosis and treatment.

  However, there are also major obstacles to continuous monitoring. First, it is very labor intensive to continuously monitor raw EEG signals. Secondly, small relative changes of the kind reflecting strokes are extremely difficult to observe, especially when only 10 seconds of data are presented at a time. QEEG may provide a solution to this problem, and much research is ongoing to try to determine what kind of computation represents the kind of change that reflects the stroke. However, research in this area has failed due to the presence of prominent artifacts in the EEG.

  In the scalp, EEG signals due to artifacts resulting from muscles, eye movements, slight electrical contact with electrodes, etc. can overwhelm brain signals. A skilled reader ignores these artifacts and concentrates on the no-artifact part, but QEEEG does not have such a margin and all signals are included in the calculation. As a result, QEEEG often reflects far more artifacts than it reflects brain activity. Thus, of course, it becomes a problem when generating graphical results, in which case a skilled reader will again be able to identify patterns resulting from brain activity. However, this is a very big problem when individual values are calculated for diagnostic purposes. For this reason, researchers often select parts that are relatively free of artifacts to perform calculations, but of course this is not clinically effective.

  Thus, there is a need for a QEEG that includes a complete signal, but with significantly reduced artifacts, especially in a clinical environment.

  The solution is to computationally remove many of the artifacts present in the recording prior to QEEG processing. In this way, the signal-to-noise ratio can be dramatically improved, and the resulting calculated value of QEEG reflects cerebral activity. At this point, it is possible to determine what kind of QEEG is useful for diagnosis and use it clinically.

  There has been research and debate in the field that the calculated value of EEG (QEEG) may be used to predict clinical symptoms of stroke.

  One of the major problems with predicting clinical symptoms of stroke using QEEG was that artifacts mixed in with cerebral signals resulted in unreliable quantitative values. The present invention achieves the level of artifact reduction currently practiced by QEEG with continuous monitoring.

  As an example of a stroke diagnosis, a physician can initiate continuous monitoring of one or more QEEEG parameters that have been determined to be symptomatic. Given an established baseline, the physician will set ranges for these parameters, and if the QEEG falls outside these ranges, the staff will be warned of the possibility of a stroke. In a more automated embodiment, the system can determine a baseline and automatically set a range, or an intelligent system such as a neural network can be used to determine a set of changes that represent QEEG and stroke. . A stroke is just one example, and many other symptoms that affect cerebral activity can be diagnosed in this way.

It is an image of quantitative EEG. It is a figure of the calculation system of quantitative EEG. It is a map of electrode arrangement | positioning of EEG. It is a detailed map of the electrode arrangement | positioning of EEG. It is a figure of a CZ standard montage. FIG. 5 is an EEG recording including seizures, muscle artifacts, and eye movement artifacts. FIG. 7 is a diagram of the EEG recording of FIG. 6 with muscle artifacts removed. FIG. 8 is a diagram of the EEG recording of FIG. 7 with eye movement artifacts removed. It is a flowchart of the calculation method of quantitative EEG. It is a flowchart of the calculation method of quantitative EEG. It is a figure of the calculation system of quantitative EEG.

  An image 100 of quantitative EEG (“qEEG”) is shown in FIG. The method and system can generate qEEGs from artifact-reduced EEG records without removing portions of the EEG records to avoid artifacts affecting the qEEGs.

  FIG. 2 shows a quantitative EEG calculation system 20. The patient 15 wears an electrode cap 31 that is attached to the patient's head, which is composed of a plurality of electrodes 35a to 35c. A processor analyzes the signals from the electrodes 35 and comprises wires 38 from the electrodes 35 connected to the EEG device elements 40 that can be used to generate EEG records 51 and qEEGs and display them on the display 50. . The electrodes utilized in the present invention are described in more detail in Wilson et al., US Pat. No. 8,121,141, “Method and Apparatus for Rapid Pressing of EEG Electrodes”, which is incorporated herein by reference in its entirety. EEG is optimized for automated artifact filtering. The EEG record is then processed using a neural network algorithm to produce a processed EEG record that is used to generate qEEG.

  Additional explanation of the analysis of EEG records can be found in Wilson et al., US patent application Ser. No. 13 / 620,855, “Method and System for Analyzing EEG Records,” filed September 15, 2012, which This is incorporated herein by reference in its entirety.

  The patient has a plurality of electrodes that are worn on the patient's head by wiring from electrodes connected to an amplifier that amplifies the signal to the processor, and the processor analyzes the signal from the electrodes and creates an EEG recording. Used to. The brain generates different signals at various points on the patient's head. As shown in FIGS. 3 and 4, a plurality of electrodes are placed on the patient's head. The CZ site is the center. For example, Fp1 in FIG. 4 is represented by channels FP1-F3 in FIG. The number of electrodes determines the number of EEG channels. The higher the number of channels, the more detailed the brain activity of the patient. Preferably, each amplifier 42 of the EEG device element 40 corresponds to two electrodes 35 that are worn on the head of the patient 15. The output from the EEG device element 40 is the difference in electrical activity detected by the two electrodes. The closer the electrode pairs are to each other, the smaller the difference in brain waves recorded by the EEG device element 40, so the placement of each electrode is important for EEG recording.

  EEG is optimized for automatic artifact filtering. The EEG record is then processed using a neural network algorithm to produce a processed EEG record that is analyzed for display. During acquisition of EEG records, the processing engine continuously analyzes the EEG waveform to determine the presence of most types of electrode artifacts for each channel. Similar to a human reader, the processing engine detects artifacts by analyzing multiple features of the EEG trajectory. Preferred artifact detection is independent of impedance checking. During acquisition, the process monitors the receive channel looking for electrode artifacts. Once detected, the artifact is automatically removed from the seizure detection process and optionally also removed from the trending display. As a result, a much higher level of seizure detection accuracy is achieved compared to previous products, and the trend is easier to read.

  Algorithms that remove artifacts from the EEG typically use blind source separation (BSS) algorithms such as CCA (canonical correlation analysis) and ICA (independent component analysis) to convert the signal from one set of channels into one set. Convert to component wave or “source”.

  In one example, an algorithm called BSS-CCA is used to remove the effect of muscle activity from the EEG. When using algorithms for recorded montages, optimal results are often not obtained. In this case it is usually best to use a montage where the reference electrode is one such as a parietal electrode such as CZ in the international 10-20 standard. According to this algorithm, the recorded montage is first converted to a CZ reference montage before artifact removal. If the CZ indicates that the signal is not the optimal choice, the algorithm refers to the list of available reference electrodes to find the appropriate electrode.

  BSS-CCA can be performed directly in the user selection montage. However, this has two problems. First, this requires performing an expensive artifact removal process at each montage selected for viewing by the user. Second, artifact removal varies from montage to montage and is optimal only when the user selects a reference montage using the optimal criteria. This is not a suitable solution because the montage required to view the EEG often does not match the optimal montage for artifact removal.

  FIGS. 5-8 illustrate a method of removing artifacts from the EEG signal to more clearly present the brain's true activity to the reader. FIG. 6 shows an EEG recording 4000 that includes seizures, muscle artifacts, and eye movement artifacts. FIG. 7 shows the EEG recording 5000 of FIG. 6 with muscle artifacts removed. FIG. 8 is a diagram illustrating the EEG recording 6000 of FIG. 7 with eye movement artifacts removed.

Various trends of EEG recordings are generated by the processing engine. The various trends, seizures probability trends, rhythm spectrogram, left hemisphere trend, rhythm spectrogram, right brain hemisphere trends, FT spectrogram left hemisphere trend, FFT spectrogram right hemisphere trends, asymmetric relative spectrogram trends, asymmetric absolute index trend, aEEG trend, inhibiting moiety is a left hemisphere and right hemisphere trends.

  Rhythm spectrograms show seizure progress in a single image. The rhythm spectrogram measures the amount of rhythm present at each frequency in the EEG recording.

  The seizure probability trend indicates the calculated probability of seizure activity over a period of time. The seizure probability trend indicates the duration of the detected seizure and suggests a recording area that does not reach the seizure detection cutoff frequency but should still be referenced. When displayed with other trends, seizure probability trends provide a comprehensive view of quantitative changes in the EEG.

  In FIG. 9, the whole calculation method of quantitative EEG is indicated by 600. At block 601, an EEG signal is generated from an EEG device comprising a plurality of electrodes, an amplifier, and a processor. At block 602, the EEG signal is continuously processed for artifact reduction to produce a processed EEG record. At block 601, a quantitative EEG is calculated from the processed EEG record. Preferably, fast EFT signal processing is used to calculate the quantitative EEG. The type of artifact to be reduced is selected from the group consisting of blinking artifacts, muscle artifacts, tongue movement artifacts, mastication artifacts, and heartbeat artifacts.

  In FIG. 10, the whole calculation method of quantitative EEG is indicated by 700. At block 701, an EEG signal is generated from an EEG device comprising an electrode, an amplifier, and a processor. At block 702, the EEG signal is continuously processed for artifact reduction to generate continuous artifact reduced EEG data. At block 703, a quantitative EEG is calculated using the continuous artifact reduction EEG data in near real time. The method further includes predicting stroke based on quantitative EEG. Alternatively, the method includes utilizing quantitative EEG for seizure detection.

11 and 12 show a calculation system for quantitative EEG. The patient 15 wears an electrode cap 31 having a plurality of electrodes 35a to 35c attached to the patient's head by wiring 38 from the electrode 35 connected to the EEG device element 40, and the EEG device element 40 includes a processor. Having an amplifier 42 that amplifies the signal sent to computer 41, the processor can analyze the signal from electrode 35 to generate EEG records 51 and qEEG 51, which can be displayed on display 50. The CPU 41 includes a software program for a neural network algorithm and a software program for a qEEG engine. As shown in FIG. 12, the artifact reduction engine, the qEEG engine 47, the microprocessor 44, the memory 42, the memory controller 43, and the I / O 48 are components of the EEG device 40. EEG is optimized for automatic artifact filtering. The EEG record is then processed using a neural network algorithm to produce a processed EEG record that is analyzed for display.