CN113576494A

CN113576494A - Electroencephalogram signal processing method and device and computer readable storage medium

Info

Publication number: CN113576494A
Application number: CN202110857473.3A
Authority: CN
Inventors: 杨小枫; 李晓楠; 闫佳庆; 任国平; 蒋忱希; 邢悦; 王玮; 李东红; 王丹; 程莉鹏; 王娇阳; 赖焕玲; 何世培
Original assignee: Bioisland Laboratory
Current assignee: Bioisland Laboratory
Priority date: 2021-07-28
Filing date: 2021-07-28
Publication date: 2021-11-02
Anticipated expiration: 2041-07-28
Also published as: CN113576494B

Abstract

Disclosed herein are a method of mapping a high-frequency oscillation conduction network, a brain electrical signal processing method, an apparatus, and a computer-readable storage medium. The method for drawing the high-frequency oscillation conduction network comprises the following steps: acquiring high-frequency oscillation events in the electroencephalogram signals of one or more time periods; and, mapping a high frequency oscillation conduction network based on the high frequency oscillation events, wherein mapping the high frequency oscillation conduction network comprises: randomly pairing all leads, each paired lead set comprising a first lead and a second lead, calculating a ratio R, and drawing the high-frequency oscillation conduction network based on the ratio R.

Description

Electroencephalogram signal processing method and device and computer readable storage medium

Technical Field

The present disclosure relates to the field of electroencephalogram signal detection technologies, and in particular, to a conduction network drawing method, an electroencephalogram signal processing apparatus, and a computer-readable storage medium based on high-frequency oscillation detection.

Background

In recent years, higher frequency bands such as High Frequency Oscillation (HFO) in intracranial brain electrical signals have received increasing attention and are considered as reliable biomarkers for epilepsy. The definition of the high-frequency oscillation is not uniformly defined at present, and the accepted standard is four continuous oscillations which are obviously higher than a base line after the band-pass filtering of 80-500 Hz. According to the frequency band, it can be further divided into Ripple (Ripple, 80-200 Hz or 80-250 Hz) and Fast Ripple (Fast Ripple, 200-500 Hz or 250-500 Hz). The high frequency oscillation is more in the initial area of the epileptic seizure, and better prognosis can be obtained by cutting off the brain area generating more high frequency oscillation.

Recently, researches suggest that high-frequency oscillation exists in network conduction and is closely related to an epilepsy-causing network, so that the high-frequency oscillation possibly participates in formation of the epilepsy-causing network and reflects excitability of the network in real time. Therefore, the high-frequency oscillation can assist in positioning the epileptogenic region and improve the curative effect of the operation. However, since the concept of the high-frequency oscillation conduction network is started later, there are few reported methods for drawing the high-frequency oscillation conduction network. At present, the granger causal analysis, the phase delay index and the like which are conventionally used abroad draw a network aiming at signals of a high-frequency band instead of high-frequency oscillation, so that the drawn network does not pay attention to the dynamic change of the high-frequency oscillation, and the evolution of the seizure-causing network is rarely understood.

Therefore, a method for accurately and rapidly drawing a high-frequency oscillation conduction network and locating key nodes in the network to find an epileptogenic network is needed to assist in excising or destroying an epileptogenic region of an epileptic patient and predict the long-term postoperative prognosis of the patient.

Disclosure of Invention

In order to solve one of the above technical problems in the prior art, the present disclosure provides an electroencephalogram signal processing method and apparatus based on high-frequency oscillation detection, and a computer-readable storage medium.

According to one aspect of the present disclosure, there is provided a method of mapping a high frequency oscillation conducting network, the method comprising the steps of: acquiring high-frequency oscillation events in the electroencephalogram signals of one or more time periods; and, mapping a high frequency oscillation conduction network based on the high frequency oscillation events, wherein mapping the high frequency oscillation conduction network comprises: randomly pairing all leads, wherein each paired lead set comprises a first lead and a second lead, calculating a ratio R according to the following formula,

plotting the high-frequency oscillation conducting network based on the ratio R.

According to another aspect of the present disclosure, there is provided a method for brain electrical signal processing, the method comprising the steps of: collecting electroencephalogram signals of a subject; selecting one or more time-period electroencephalogram signals from the acquired electroencephalogram signals; acquiring high-frequency oscillation events in the electroencephalogram signals of the one or more time periods; and, mapping a high frequency oscillation conduction network based on the high frequency oscillation events, wherein mapping the high frequency oscillation conduction network comprises: randomly pairing leads, wherein each paired lead set comprises a first lead and a second lead, calculating a ratio R according to the following formula,

According to some embodiments of the present disclosure, the step of acquiring high-frequency oscillation events in the electroencephalogram signal of one or more time periods may include arranging absolute amplitudes of high-frequency oscillation waves in the electroencephalogram signal from small to large to obtain a maximum peak point distribution curve, determining a turning point on the maximum peak point distribution curve, and taking an average value of the absolute amplitudes of all oscillation waves before the turning point as a baseline. According to some embodiments of the present disclosure, the method may include determining a dynamic baseline of the brain electrical signal for the one or more time periods using a moving time window over the one or more time periods.

In some embodiments of the present disclosure, the step of acquiring high-frequency oscillation events in one or more time periods of electroencephalogram signals may include acquiring high-frequency oscillation events in electroencephalogram signals for more than 5 minutes, for example, high-frequency oscillation events in electroencephalogram signals for 5 minutes, 10 minutes, 15 minutes, 20 minutes, 25 minutes, 30 minutes, 1 hour, 2 hours, or even longer time periods may be acquired. It will be understood by those skilled in the art that the time period may be set according to the condition of brain wave conduction of the subject and the actual needs in the implementation process, and the present disclosure does not limit this.

According to some embodiments of the present disclosure, when an electroencephalogram signal is acquired by using a grid-shaped or sheet-shaped electrode, after 80-200 Hz filtering is performed, when there are 8 or more continuous peak point amplitudes higher than 3.5 times of the standard deviation of the baseline, and 6 or more peak point amplitudes higher than 9 times of the SD of the baseline, the high-frequency oscillation event is determined to be a ripple. According to some embodiments of the present disclosure, when electroencephalogram signals are acquired by using a grid-shaped or sheet-shaped electrode, after 200-500 Hz filtering, when there are 8 or more continuous peak point amplitudes higher than 3 times of standard deviation of the baseline, and 4 or more peak points higher than 10.5 times of standard deviation of the baseline, the high-frequency oscillation event is determined to be a rapid ripple.

According to some embodiments of the present disclosure, when electroencephalogram signals are acquired by using a stereospecific electrode, after 80-200 Hz filtering, when there are 8 or more continuous peak point amplitudes higher than 3 times of the standard deviation of the baseline, and 6 or more peak points higher than 10 times of the standard deviation of the baseline, the high-frequency oscillation event is determined to be a ripple. According to some embodiments of the present disclosure, when electroencephalogram signals are acquired by using a grid-shaped or sheet-shaped electrode, after 200-500 Hz filtering is performed, when there are 8 or more continuous peak point amplitudes higher than 3 times of standard deviation of the baseline, and 6 or more continuous peak point amplitudes higher than 9.5 times of standard deviation of the baseline, the high-frequency oscillation event is determined to be a rapid ripple.

According to some embodiments of the present disclosure, the step of acquiring high frequency oscillation events in the brain electrical signal of one or more time periods further comprises removing false high frequency oscillation events. According to some embodiments of the present disclosure, based on a frequency spectrum density map drawn by a frequency spectrum phase space reconstruction method, a frequency offset energy difference is detected, frequency-dependent frequency offset energy difference values of a frequency band in which the high-frequency oscillation event is located are accumulated, and if an obtained value is less than or equal to 0, the corresponding high-frequency oscillation event is determined to be a false high-frequency oscillation event, and is removed. According to some embodiments of the disclosure, the false high frequency oscillation event is caused by a gibbs effect.

According to some embodiments of the present disclosure, the high frequency oscillation event occurring on the second lead associated with the first lead means that a high frequency oscillation event also occurs on the second lead within a time period t after the high frequency oscillation event occurs on the first lead. According to some embodiments of the present disclosure, the time period t is 100 to 300 ms. According to some embodiments of the present disclosure, the time period t is 100 to 250 ms. According to some embodiments of the disclosure, the time period t is 100, 150, 200, 250 or 300 ms. It will be understood by those skilled in the art that the time period t may be set according to the speed of brain wave conduction of the subject in practice, and the present disclosure is not limited thereto.

According to some embodiments of the disclosure, the plotting the high-frequency oscillation conducting network based on the ratio R may comprise: ranking the leads based on the magnitude of the ratio R. According to some embodiments of the present disclosure, in all paired lead groups with one lead as the first lead, when more than 2 ratios R in all calculated ratios R are greater than or equal to 0.5, it is determined that the lead actively conducts a high-frequency oscillation event. According to some embodiments of the present disclosure, in all paired lead groups with one lead as the second lead, when the ratio R ≧ 0.5 is less than 2 of all the calculated ratios R, the lead is judged not to receive the high-frequency oscillation event.

According to some embodiments of the present disclosure, mapping the high frequency oscillation conduction network based on the ratio R includes determining that a lead is a level 1 node when the lead is actively conducting high frequency oscillation events and is not receiving high frequency oscillation events. According to some embodiments of the present disclosure, when a first lead in the paired lead set is determined as the level 1 node, if the ratio R of the paired lead set is greater than or equal to 0.5 and a second lead actively conducts a high-frequency oscillation event, the second lead of the paired lead set is determined as the level 2 node, otherwise, the second lead is determined as the level 3 node. According to some embodiments of the present disclosure, when the first lead in the paired lead set is determined as the level 2 node, if the ratio R of the paired lead set is greater than or equal to 0.5, the second lead of the paired lead set is determined as the level 3 node. According to some embodiments of the present disclosure, in the high frequency oscillation conduction network, the high frequency oscillation event is conducted from the level 1 node to the level 2 node, and from the level 2 node to the level 3 node. In some embodiments, the high frequency oscillation event may also be conducted from the level 1 node to the level 3 node.

For example, the ratio R is the active conduction probability R for a first lead in the pair-lead set relative to a second lead₁For the second lead, it is its passive reception probability R relative to the first lead₂. After all leads are randomly paired, at least 2 active conduction probabilities R for a lead relative to other leads if the lead is the first lead₁If the lead is more than or equal to 0.5, the lead has active conduction capability; at most 2 passive reception probabilities R relative to other leads if it is the second lead₂And more than or equal to 0.5, the passive receiving capability of the lead does not exist. In some embodiments, a lead is determined to be a level 1 node when only active conduction capability exists and no passive reception capability exists for the lead. In some embodiments, the passive reception probability R when a lead is relative to a level 1 node₂And more than or equal to 0.5, and judging the lead as a level 2 node when the active conduction capability exists, otherwise, judging the lead as a level 3 node. In some embodiments, the probability of passive reception R when a lead is relative to a level 2 node₂And (5) not less than 0.5, and judging the lead as a level 3 node whether the active conduction capability exists or not.

According to yet another aspect of the present disclosure, there is provided an apparatus for mapping a high frequency oscillation conducting network, the apparatus comprising: the first calculation module is used for acquiring high-frequency oscillation events in the electroencephalogram signals of the one or more time periods; a second calculation module for mapping a high frequency oscillation conduction network based on the high frequency oscillation events, wherein the second calculation module is used for randomly pairing leads, each paired lead group comprises a first lead and a second lead, and the ratio R is calculated according to the following formula,

According to yet another aspect of the present disclosure, there is provided an apparatus for brain electrical signal processing, the apparatus comprising: the acquisition module is used for acquiring electroencephalogram signals of a subject; the processing module is used for selecting one or more time-period electroencephalogram signals from the acquired electroencephalogram signals; the first calculation module is used for acquiring high-frequency oscillation events in the electroencephalogram signals of the one or more time periods; a second calculation module for mapping a high frequency oscillation conduction network based on the high frequency oscillation events, wherein the second calculation module is configured to randomly pair the leads, each paired lead comprising a first lead and a second lead, calculate a ratio R according to the following formula,

According to some embodiments of the present disclosure, the first calculating module may include a baseline determining module, where the baseline determining module is configured to arrange absolute amplitudes of high-frequency oscillatory waves in the electroencephalogram signal from small to large to obtain a maximum peak point distribution curve, determine a turning point on the maximum peak point distribution curve, and use an average value of the absolute amplitudes of all oscillatory waves before the turning point as a baseline. According to some embodiments of the present disclosure, the baseline determination module may be configured to determine the dynamic baseline of the brain electrical signal for the one or more time periods using a moving time window over the one or more time periods.

According to some embodiments of the present disclosure, the first calculation module may further comprise a traversal module for determining parameters for detecting the high-frequency oscillation event, wherein the parameters include a low threshold, a high threshold, and a number of high-frequency oscillation peak points satisfying the high threshold.

According to some embodiments of the present disclosure, the high frequency oscillation event occurring on the second lead associated with the first lead means that a high frequency oscillation event also occurs on the second lead within a time period t after the high frequency oscillation event occurs on the first lead. According to some embodiments of the present disclosure, the time period t is 100 to 300 ms.

According to some embodiments of the present disclosure, the second calculation module may further include a decision module for ranking the leads based on the magnitude of the ratio R. According to some embodiments of the present disclosure, in all paired lead groups in which one lead is used as the first lead, when more than 2 ratios R in all calculated ratios R are greater than or equal to 0.5, the determination module may determine that the lead actively conducts a high-frequency oscillation event. According to some embodiments of the present disclosure, in all paired lead groups in which one lead is used as the second lead, when the ratio R of less than 2 of all calculated ratios R is greater than or equal to 0.5, the determination module may determine that the lead does not receive the high-frequency oscillation event.

According to some embodiments of the present disclosure, when a lead actively conducts high frequency oscillation events and does not receive high frequency oscillation events, then the decision module may be configured to decide that the lead is a level 1 node. According to some embodiments of the present disclosure, when the first lead in the paired lead set is determined as the level 1 node, if the ratio R of the paired lead set is greater than or equal to 0.5 and the second lead actively conducts the high-frequency oscillation event, the determining module may be configured to determine that the second lead of the paired lead set is a level 2 node, otherwise, determine that the second lead is a level 3 node. According to some embodiments of the present disclosure, when the first lead in the paired lead set is determined as the level 2 node, if the ratio R of the paired lead set is greater than or equal to 0.5, the determination module may be configured to determine that the second lead of the paired lead set is a level 3 node. According to some embodiments of the present disclosure, in the high frequency oscillation conduction network, the high frequency oscillation event may be determined to be conducted from the level 1 node to the level 2 node and from the level 2 node to the level 3 node.

According to some embodiments of the present disclosure, the apparatus of the present disclosure may further comprise a fusion module for fusing the high-frequency oscillation conduction network to a three-dimensional brain model of the subject. According to some embodiments of the present disclosure, the three-dimensional brain model may be obtained by Magnetic Resonance Imaging (MRI) techniques. According to some embodiments of the present disclosure, the position of each electrode point is displayed on the three-dimensional brain model based on preset coordinates of an implanted electrode, a CT scan image after the electrode is implanted, and other data. According to some embodiments of the present disclosure, a distribution of high frequency oscillation events is also displayed on the three-dimensional brain model. According to some embodiments of the present disclosure, the apparatus may further comprise a monitoring module for monitoring the number of high frequency oscillation events and/or changes in the high frequency oscillation conduction network.

According to still another aspect of the present disclosure, there is provided a computer-readable storage medium having stored thereon a program which, when processed and executed, may implement the method of drawing a high-frequency oscillation conductive network or the method for brain electrical signal processing of the present disclosure.

According to still another aspect of the present disclosure, there is provided a use of the above-mentioned method for mapping a high-frequency oscillation conduction network of the present disclosure in the preparation of an apparatus for mapping a high-frequency oscillation conduction network, or a use of the above-mentioned method for electroencephalogram signal processing in the preparation of an apparatus for electroencephalogram signal processing.

According to yet another aspect of the present disclosure, a surgical system is provided that includes the apparatus or computer-readable storage medium of the present disclosure. According to some embodiments, the surgical detection system may be used to perform the above-described method of mapping a high-frequency oscillatory conductive network or the above-described method for electroencephalogram signal processing.

According to still another aspect of the present disclosure, a high-frequency oscillation conduction network model is provided, which is mapped by the method for mapping a high-frequency oscillation conduction network or the method for brain electrical signal processing of the present disclosure.

Compared with the prior art, the method is not only suitable for monitoring the sickroom or the clinic before the operation, but also suitable for real-time monitoring in the operation, so that the problem that the electroencephalogram signal is monitored for a long time before the operation so as to consume a large amount of manpower, material resources and financial resources is avoided. The method has the advantages that intracranial high-frequency oscillation events are detected in real time in the operation, and key nodes on an epilepsy-causing area and an epilepsy-causing network can be accurately positioned by drawing a high-frequency oscillation conduction network. The method comprises the steps of finding key nodes in a conduction network, fusing the high-frequency oscillation conduction network with an individual brain model of a subject, and correspondingly depicting and marking the suspicious seizure area positions and the key nodes. Therefore, the distribution and the conduction condition of the high-frequency oscillation can be known more intuitively, so that the planning and the execution of an operation plan are facilitated, the planning of an operation scheme is guided, the dynamic change of a network is monitored in real time to evaluate the curative effect of the operation, the success rate of the operation is improved, and the method has certain guiding and predicting significance on the postoperative prognosis of epileptics.

Drawings

In order that the disclosure may be more readily and clearly understood, reference is now made to the following detailed description of the disclosure taken in conjunction with the accompanying drawings, in which

Fig. 1 shows a flow diagram of a method of mapping a high frequency oscillation conducting network according to one embodiment of the present disclosure.

Fig. 2 shows a flow diagram of a brain electrical signal processing method according to an embodiment of the present disclosure.

Fig. 3 shows a schematic flow chart of detection implemented in epileptic surgery by the electroencephalogram signal processing method according to one embodiment of the present disclosure.

Fig. 4 illustrates a high frequency oscillation detection system according to one embodiment of the present disclosure. (A) An operation interface of the high-frequency oscillation detection system; (B) is the result of the analysis; (C) the detection report output by the high-frequency oscillation detection system comprises high-frequency analysis results required by the epilepsy preoperative evaluation, such as high-frequency oscillation sequencing, high-frequency oscillation region distribution and high-frequency oscillation brain region distribution.

FIG. 5 illustrates a schematic diagram of detecting high frequency oscillation events according to one embodiment of the present disclosure. (A) The procedure for determining high frequency oscillation events is shown schematically. (B) The upper graph is a Peak Point Distribution Curve (PPDC) of the ripple maximum wave made by 80-200 Hz band-pass filtering, and the lower graph is a rapid PPDC of the ripple made by 200-500 Hz band-pass filtering. (C) PPDC pattern diagram. Curves (i), (ii) and (iii) represent the lowest, medium and highest amount of high frequency activity contained in the electroencephalogram signal, respectively.

FIG. 6 shows a schematic diagram of manual identification and automatic detection of ripple according to one embodiment of the present disclosure. (A) Six seconds of raw electroencephalographic data. (B) Enlargement of data in dashed box in a. (C, D) manually identified ripples and fast ripples, respectively, after filtering. (E, F) ripple and Fast Ripple (FR) detected based on the automatic detection method of the present disclosure, respectively. (G, H) is a time-frequency spectrum calculated by wavelet analysis.

Fig. 7 shows a schematic diagram of distinguishing between true high frequencies and false high frequencies according to one embodiment of the present disclosure. (A) True high frequency oscillation events, (B) false high frequency oscillations due to the gibbs effect.

FIG. 8 illustrates an analytical schematic of a false dither event according to one embodiment of the present disclosure. (A, C) are the power spectral density map of true ripples and the Morlet wavelet analysis spectrogram, respectively. (B, D) are the power spectral density map of true fast ripples and the Morlet wavelet analysis spectrogram, respectively. (E) Representing the corresponding energy accumulation index of high frequency oscillation events within one lead.

Fig. 9 shows a schematic diagram of a time delay method for drawing a high-frequency oscillation conducting network according to an embodiment of the present disclosure. (A) A high frequency oscillating conducting network plotted against 250ms delay is illustrated, in which

level

1, 2, 3 nodes determined according to the method of the present disclosure are shown. (B) The calculation principle is shown by way of example. (C) Is an example based on SEEG electroencephalogram data.

Figure 10 illustrates the ripple and rapid ripple and brain pattern for two surgically resected patients. (A) An example of a patient with good surgical prognosis is shown. (B) An example of a patient with poor surgical prognosis is shown.

FIG. 11 is a graph illustrating the change in the high frequency oscillatory conduction network of a patient after a thermal coagulation failure. (A) The level 1 and level 2 nodes, leads 3-6 of electrode A and leads 1-4 of electrode E, as well as the high frequency oscillation events before and after thermal coagulation damage, determined based on the high frequency oscillation conduction network, are shown. (B) The broadband electroencephalogram recording conditions before and after the thermal coagulation damage are displayed. (C) A schematic diagram of automatically detected high frequency oscillations is schematically shown. (D) The upper graph shows the variation of the high-frequency oscillation quantity synchronously displayed on the brain model of the patient before and after the damage; the lower graph shows the dynamic change of the high frequency oscillating conducting network before and after the failure.

Fig. 12 shows a schematic diagram of a high frequency oscillation conduction network and its key nodes fused with brain models according to an embodiment of the present disclosure.

FIG. 13 shows a brain electrical signal processing apparatus 300 according to one embodiment of the present disclosure.

Detailed Description

In order to make the objects, technical solutions and advantages of the present disclosure more apparent, the present disclosure is further described in detail below with reference to the following embodiments. The specific embodiments described herein are merely illustrative of the disclosure and do not constitute any limitation on the disclosure. Moreover, in the following description, descriptions of well-known structures and techniques are omitted so as to not unnecessarily obscure the concepts of the present disclosure.

All terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art unless otherwise defined. It is noted that the terms used herein should be interpreted as having a meaning that is consistent with the context of this specification and should not be interpreted in an idealized or overly formal sense.

The term "epilepsy" as used herein refers to a chronic disease resulting in transient cerebral dysfunction due to sudden abnormal firing of cerebral neurons. Many intractable epilepsy can not be cured by medicine, and the trouble caused by the disease can be eliminated only by means of operation treatment. In epileptic neurosurgery, the positioning of epileptic foci is very important. Preliminary studies have found that a High Frequency Oscillation (HFO) signal in an electroencephalogram signal can directly reflect the synchronization activity of neurons, is closely related to epileptic seizure, more accurately indicates an epileptic seizure onset region than epileptic-like discharges, and can be used as a biomarker of the epileptic seizure onset region.

For electroencephalogram oscillatory waves, the term "absolute amplitude" as used herein is calculated using the highest and lowest points of each oscillatory wave in the electroencephalogram signal as peak points.

Some block diagrams and/or flow diagrams are shown in the figures. It will be understood that some blocks of the block diagrams and/or flowchart illustrations, or combinations thereof, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the instructions, which execute via the processor, create means for implementing the functions/acts specified in the block diagrams and/or flowchart block or blocks. The techniques of this disclosure may be implemented in hardware and/or software (including firmware, microcode, etc.). Additionally, the techniques of this disclosure may take the form of a computer program product on a computer-readable storage medium having instructions stored thereon for use by or in connection with an instruction execution system.

Example 1

Fig. 1 schematically shows a flow chart of a method for mapping a high-frequency oscillation conducting network. The method for drawing the high-frequency oscillation conduction network comprises the following steps:

s101: acquiring high-frequency oscillation events in the electroencephalogram signals of one or more time periods; and

s102: drawing a high-frequency oscillation conducting network based on the high-frequency oscillation event, wherein the drawing the high-frequency oscillation conducting network comprises: randomly pairing all leads, wherein each paired lead set comprises a first lead and a second lead, calculating a ratio R according to the following formula,

The method of the embodiment can be used for acquiring high-frequency oscillation events of one or more time periods from an existing electroencephalogram signal database. For example, a patient's electroencephalographic database may be used for retrospective analysis of the condition.

In the present embodiment, electroencephalogram signals from 2 or more leads, which are randomly paired to calculate 2 or more ratios R, are acquired. And drawing the high-frequency oscillation conduction network according to the obtained ratio R.

Example 2

Fig. 2 shows a flow chart of the electroencephalogram signal processing method of the present disclosure. The electroencephalogram signal processing method comprises the following steps:

s201: collecting electroencephalogram signals of a subject;

s202: selecting one or more time-period electroencephalogram signals from the acquired electroencephalogram signals;

s203: acquiring high-frequency oscillation events in the electroencephalogram signals of the one or more time periods;

s204: drawing a high frequency oscillation conduction network based on the high frequency oscillation events, wherein the leads are randomly paired, each paired lead comprises a first lead and a second lead, the ratio R is calculated according to the following formula,

drawing a high-frequency oscillation conducting network based on the ratio R.

Example 3

FIG. 3 illustrates a flow chart of a method for intra-operative real-time electroencephalogram signal processing to particularly illustrate the method of electroencephalogram signal processing of the present disclosure. The method can acquire the electroencephalogram signal of a subject through an electroencephalogram signal detection device, determine a high-frequency oscillation event according to the acquired electroencephalogram signal, draw a high-frequency oscillation conduction network based on the detected high-frequency oscillation event, and judge a focus point through fusion with a brain model.

S201: collecting the EEG signal of the subject.

Specifically, the intracranial broadband electroencephalogram signals of the subject are collected in real time in the operation process through the electrodes. Preferably, the electroencephalogram signals are collected under the condition that the electromagnetic interference of the surrounding environment is small. A video electroencephalograph with a higher signal-to-noise ratio amplifier can be used to acquire electroencephalographic signals.

The electroencephalogram signals may be acquired through subdural electrodes, stereotactic electrodes, or electroencephalogram electrodes of the cerebral cortex. For example, as shown in FIG. 3, intracranial brain electrical signals are acquired in real time through SEEG/Depth/ECoG electrodes. Optionally, the electroencephalogram signal with the frequency of 1000-2000Hz or above 2000Hz is collected.

S202: and randomly selecting one or more electroencephalogram signals with less interference in a time period from the acquired electroencephalogram signals.

Specifically, electroencephalogram signal data may be randomly selected for a period of time (e.g., 5 minutes) on an electroencephalograph. Optionally, when the electroencephalogram signal data are selected, data segments with less electromyography, electrocardio and electromagnetic interference are selected. As shown in fig. 3, brain electrical signals for a 5 minute slow wave sleep period were acquired for patients using 7 subdural electrodes (including 70 leads) or 24 segg electrodes (including 48 leads).

And editing the electroencephalogram signals. The electroencephalogram signals can be exported in files in EDF format. Then double guides are edited and stored as MAT format data. It should be noted that some electroencephalographs can directly derive the electroencephalogram data with double leads already compiled, so that double leads do not need to be edited.

The electroencephalogram signal acquisition during the operation may be influenced by anesthesia. Although it has been found that monitoring high frequency oscillations under intraoperative anesthesia can still accurately locate epileptogenic regions, it is suggested that intracranial brain electrical data be randomly selected for a period of time (e.g., 5 minutes) prior to anesthesia for the following high frequency oscillation analysis. Or, before anesthesia, intracranial electroencephalogram data of a period of time can be selected as a contrast, and compared with the intracranial electroencephalogram data selected after anesthesia, so as to judge whether the anesthesia can affect the whole high-frequency oscillation acquisition and analysis result.

The EEG signals can be selected from the group consisting of an inter-seizure period (e.g., waking period or sleeping period), a seizure period, and a pre-seizure and post-seizure period.

S203: and acquiring high-frequency oscillation events in the electroencephalogram signals of the one or more time periods.

When detecting ripple, a band-pass filtering of 80-200 Hz can be performed. In some embodiments, when fast ripple is detected, a 200-500 Hz bandpass filtering may be performed. Then, based on the maximum distribution peak point principle, a dynamic baseline is determined. Further, the dynamic baseline may be determined by using a moving time window. As shown in fig. 3, a 5 second moving time window is used to calculate the dynamic baseline based on the maximum distribution peak point principle.

FIG. 4A illustrates an operating system interface of the brain electrical signal processing device. And on the operating system interface, performing band-pass filtering on the stored electroencephalogram signal data by using a finite pulse band-pass filter. A baseline of the brain electrical signal is determined for a 1s time period in a 5 second time window. Firstly, all electroencephalogram signals fo (t) in the time period are obtained, and after band-pass filtering (80-200 Hz and 200-500 Hz), the filtered electroencephalogram signals f (x) are obtained. The filter is set to a range of HFO frequencies for which a baseline calculation is required.

Deriving the filtered signal f (t) to obtain f' (t), then solving the equation: f' (t) ═ 0.

Assuming the root of the equation is r, the extreme point (i.e., peak point) of the signal f' (t), extreme (r), can be obtained.

The obtained extreme points are sorted from small to large according to absolute values, and the sequential description of the signal amplitude distribution can be obtained:

Order(r)＝sort(|Extremum(r)|)。

FIG. 5 illustrates the detection of a high frequency oscillation event by finding a baseline peak point by calculating PPDC. As shown by PPDC in fig. 5A, the high frequency components (high frequency oscillation events and high frequency noise) in the signal are no more than 40% of the total signal, and the extremum point ordering signal order (r) exhibits a linear region in the range of 20% to 60%. Therefore, linear fitting is carried out on the range of 20% -60% of the order (r) signal to obtain a straight line:

y＝a×r﹢b。

since the high frequency components do not participate in the calculation of the line parameters, the parameters a and b of the line will have a strong correlation with the baseline of the high frequency signal. From the linear parameters a and b, it can be obtained that when r takes a value of 100% (set as r0), the value of y has a fixed coefficient relationship with the signal baseline value, and the coefficient is set as G, and finally the baseline part of the high-frequency oscillation can be estimated as:

Base＝G×(a×r0﹢b)

g can be obtained through electroencephalogram data testing, and will typically be related to the hardware settings of the electroencephalogram acquisition device.

The dotted line portion in the PPDC curve in fig. 5A is the obtained fitted straight line. And selecting a first peak point with the absolute amplitude higher than 5% of the straight line in the PPDC curve as a turning point, wherein the peak point before the turning point is a baseline peak point. The mean absolute amplitude of all baseline peak points is taken as the peak baseline (not shown in the figure). Through similar steps, a moving time window (e.g., a 5 second moving time window) may be used to obtain a dynamic baseline as shown in FIG. 5B.

After the dynamic baseline is determined, different frequency bands of the high frequency oscillation, namely ripple and fast ripple, are detected respectively based on traversing the determined parameters. Such as the parameter optimization step shown in fig. 3. The parameters may include a low threshold, a high threshold, the number and step size of peak points satisfying the high threshold, and the like. Alternatively, if the interval between consecutive dither events is less than 25 milliseconds of ripple or less than 10 milliseconds of fast ripple, then the dither events are merged. The detection of high frequency oscillation events is specifically set forth below in connection with the different electrode types used.

For a grid-shaped or sheet-shaped electrode (for example, a subdural electrode), after filtering at 80-200 Hz, when the amplitude of at least 8 continuous wave peak points is higher than 3.5 times Standard Deviation (SD) of a base line, and the amplitude of at least 6 wave peak points is higher than 9 times SD of the base line, judging the oscillation as ripple; after filtering at 200-500 Hz, when the amplitude of at least 8 continuous wave peak points is higher than 3 SD times of the baseline, and at least 4 wave peak points are higher than 10.5 SD times of the baseline, the oscillation is judged to be rapid ripple.

For a stereospecific electrode (for example, SEEG electrode), after filtering at 80-200 Hz, when the amplitude of at least 8 continuous wave peak points is higher than 3 times SD of a base line, and at least 6 wave peak points are higher than 10 times SD of the base line, judging the oscillation as ripple; after filtering at 200-500 Hz, when the amplitude of at least 8 continuous peak points is higher than 3 times SD of the base line, and at least 6 of the amplitude are higher than 9.5 times SD of the base line, the oscillation is judged to be rapid ripple.

Optionally, the frequency offset energy difference is detected based on a spectral density map drawn by a spectral phase space reconstruction method to remove the false high frequency oscillation event. The false high frequency oscillation event may be caused by the gibbs effect. The analysis principle for removing false dither events is illustrated in fig. 8. Specifically, the false high frequency oscillations formed by filtering appear as "pagoda" or "mountain peaks" on the spectrogram of Morlet wavelet analysis, and are distinguished from the "islands" or "water drops" of true high frequency oscillations. When the time-frequency spectrogram corresponding to the high-frequency oscillation is in a pagoda shape or a mountain peak shape, marking the time-frequency spectrogram as false high-frequency oscillation; when the time-frequency spectrogram corresponding to the high-frequency oscillation shows a free island shape or a free water drop shape, the mark is true high-frequency oscillation. And selecting a middle point of each high-frequency oscillation event time, measuring the spectrum energy of each frequency band through Morlet wavelet analysis, and making a spectrum density graph. On the spectral density map, the energy of the true high-frequency oscillations increases significantly in the measured frequency range, but the false high-frequency oscillations do not show a significant energy increase. And detecting the true and false high frequency by using a spectral phase space reconstruction method according to the phenomenon that the energy of the frequency band corresponding to the frequency band density graph where the true high frequency oscillation is located is increased.

In determining true and false high frequency oscillation events, a frequency offset energy difference calculation formula, Power (f) -Power (f- Δ f), where f is frequency, may be used. When the detected dither event is a false dither event, the energy at the higher frequency will be lower than the energy at the lower frequency, presenting a tendency for the energy to decrease as the frequency increases, i.e. Power (f) -Power (f- Δ f) <0. On the contrary, when the detected HFO event is a true HFO event, the energy of the frequency band in which the high frequency oscillation event is located shows a trend of increasing compared with the energy of the previous frequency band, i.e. Power (f) -Power (f- Δ f) ≧ 0. To identify complete information, as complete a frequency offset signature as possible can be reflected by gradually increasing the Δ f value. The frequency f is increased from 80Hz to 600Hz (step size of 1Hz), and Δ f is increased from 1Hz to 200Hz (step size of 1 Hz). When Power (f) -Power (f- Δ f) <0, the point is set to 0, otherwise it is not changed. Thus, when Power (f) -Power (f- Δ f) ≦ 0, the corresponding high frequency oscillation is set to be a false high frequency oscillation event, and the removal is performed.

The limitation based on the energy of the baseline is optionally continued to be increased since the energy of the baseline may affect the detection of high frequency oscillation events. Since the energy of the high-frequency oscillation event is certainly greater than the baseline energy, if the energy difference is smaller than the baseline fluctuation energy, the frequency energy difference value should be set to 0. And finally, accumulating the frequency energy difference of the frequency band where the high-frequency oscillation is positioned, and if the value is 0, indicating that the frequency band has no energy enhancement phenomenon, namely no high-frequency oscillation event is generated.

As shown in fig. 8E, if a high frequency oscillation event occurs, energy enhancement occurs in the corresponding frequency band and the delay frequency band. And accumulating the energy of the corresponding frequency band, wherein the true high-frequency energy accumulation index (frequency offset energy accumulation value) is greater than 0. And the false high-frequency oscillation event has no energy enhancement, and the energy accumulation index is less than 0. Therefore, the true and false high-frequency oscillation events can be effectively distinguished, and the false high-frequency oscillation events can be removed.

Figure 7 shows an example of automatic detection to distinguish between a true high frequency oscillation event and a false high frequency oscillation event. In FIG. 7A, (a) is 1s of raw brain electrical data; (b) amplifying data in a virtual frame in the step (a); (c) and (d) true ripple and true fast ripple detected, respectively; (e) and (f) time-frequency spectrograms of the corresponding Morlet wavelet analysis are respectively shown, wherein the time-frequency spectrograms corresponding to the true high-frequency oscillation are all shown as free energy rising graphs. In FIG. 7B, (a) is 1s of raw brain electrical data; (b) amplifying data in a virtual frame in the step (a); (c) and (d) is the detected false ripple and true fast ripple; (e) and (f) is the time-frequency spectrum of the corresponding Morlet wavelet analysis, wherein the peak or pagoda-like energy rise is shown on the time-frequency spectrum corresponding to the false high-frequency oscillation.

A schematic diagram of manual identification and automatic ripple detection is shown in fig. 6. Fig. 6A is a 6s segment of raw electroencephalogram data (intracranial subdural electrodes, sample rate 2000Hz) in which 1s of data was randomly selected (as shown by the dashed box in the figure). Fig. 6B is an enlargement of the electroencephalogram data in the dashed-line box of fig. 6A. Figures 6C and 6D show the artificially identified ripple and the Fast Ripple (FR), respectively, after filtering. Fig. 6E and 6F show the ripple and the fast ripple detected by applying the above-mentioned automatic detection method of high frequency oscillation based on the peak point of maximum distribution after filtering, respectively. The vertical lines in fig. 6C, 6D, 6E, and 6F indicate the start and end points of the high-frequency oscillation event s, which confirms that the automatic detection method of the present disclosure can correctly detect the high-frequency oscillation event, and the result of the manual identification is consistent. Fig. 6G and 6H are time-frequency spectrograms calculated using wavelet analysis. It can be seen that the detected high-frequency oscillation events all show free-form energy rise in the video spectrogram, and are therefore true high-frequency oscillation events.

S204: and drawing the high-frequency oscillation conduction network by a time delay method.

All electrode leads are paired randomly as shown in fig. 3. For the paired leads, the ratio R of the number of high frequency oscillation events occurring on the second lead within 250ms of the occurrence of high frequency oscillation events on the first lead (i.e. the number of high frequency oscillation events associated with each other) to the total number of high frequency oscillation events occurring on the second lead is calculated. The ratio R is 0 to 1. The closer the ratio R is to 1, the more conductive the first lead to the second lead. After the ratio R is obtained through pairwise pairing and combination calculation of all leads, the conductivity of all leads can be obtained through comparison, grading is carried out, and the conduction relation is determined. It is believed that high frequency oscillation events are conducted from a more conductive lead to a less conductive lead. Thus, conduction guidance between leads can be derived from the ratio R of all lead pairs. And drawing a high-frequency oscillation conduction network according to the conductivity strength and the conduction guide of each lead. And judging key nodes in the high-frequency oscillation conduction network according to the conduction strength.

According to one embodiment of the present disclosure, there are n leads. Pairing lead 1 with lead 2, setting lead 1 as a first lead and lead 2 as a second lead, calculating the ratio of the number of high-frequency oscillation events generated by lead 2 and correlated with lead 1 to the total number of high-frequency oscillation events generated on lead 2, and obtaining the ratio R_1-2. In a similar way, the ratio of each pair of leads can be found:

R_1-2、R_1-3、R_1-4、……、R_1-n；

R_2-1、R_2-3、R_2-4、……、R_2-n；

R_3-1、R_3-2、R_3-4、……、R_3-n；

……；

R_n-1、R_n-2、R_n-3、……、R_n-(n-1)。

based on the magnitude of the above ratio R, the leads are ranked, for example, into 3 levels. The high frequency oscillation event may be conducted from the level 1 node to the level 2 node and then from the level 2 node to the level 3 node (as exemplarily shown in fig. 9). These critical nodes tend to play a significant role in initiation and diffusive propagation in high frequency oscillatory conductive network conduction. An exemplary description is provided below.

If lead 1 is taken as all ratios of the first lead, R_1-2、R_1-3、R_1-4、……、R_1-nWhen the ratio R of 2 or more than 2 in the lead 1 is more than or equal to 0.5, the lead 1 is judged to actively conduct the high-frequency oscillation event. If lead 1 is the second lead's all ratios, R_2-1、……、R_n-1When only 0, 1 or 2 ratio values R is more than or equal to 0.5, judging that lead 1 does not receive the high-frequency oscillation event. If it is judged that lead 1 is actively conducting high frequency oscillation events and is not receiving high frequency oscillation events, lead 1 is determined to be a level 1 node.

If lead 2 is the first lead's all ratios, R_2-1、R_2-3、R_2-4、……、R_2-nWhen the ratio R of 2 or more than 2 in the lead 2 is more than or equal to 0.5, the lead 2 is judged to actively conduct the high-frequency oscillation event. In the paired lead set (leads 1-2) in which lead 1 is the first lead and lead 2 is the second lead, if the ratio R is at level 1 node in lead 1_1-2≧ 0.5 and lead 2 actively conducts the high-frequency oscillation event, then lead 2 is determined as the level 2 node, otherwise lead 2 is determined as the level 3 node.

In the paired lead set (leads 2-4) in which lead 2 is the first lead and lead 4 is the second lead, when lead 2 is determined to be the level 2 node,if the ratio R of the paired pilot group_2-4And > 0.5, then lead 4 is determined to be a level 3 node.

Based on the above-described grading, it can be determined that the high frequency oscillation event is conducted from lead 1 to lead 2 and then from lead 2 to lead 4. In a similar way, all leads are ranked and the conduction network of high frequency oscillation events is determined.

Figure 9 illustrates the high frequency oscillating conducting network plotted against a 250ms delay. As shown in FIG. 9B, the leads are paired two by two, and within 250ms of the occurrence of a high frequency oscillation event on the first lead (channel 1), a high frequency oscillation event also occurs on the second lead (channel 2), and the ratio R is calculated_1-2. Calculating a plurality of ratios R for channel 1 as the first lead_1-n≧ 0.5 (not shown in the figure), and channel 1 as the second lead has no corresponding ratio R (i.e., no high-frequency oscillation event is received from the other channels), therefore channel 1 is identified as the level 1 node. In the channel 1 to channel 2 pairing group, the ratio R_1-2Not less than 0.5, and the ratio R of channel 2 to channel 3_2-3≧ 0.5 (i.e., channel 2 actively conducts a high-frequency oscillation event to channel 3), then channel 2 is determined to be the level 2 node. In the paired conducting set of channel 2 and channel 3, the ratio R_2-3≧ 0.5, lane 3 is determined as a level 3 node. Thus, the high frequency oscillation event is conducted from channel 1 to channel 2 and then to channel 3.

Optionally, the ratio R of the number of dither events for the paired lead set is used as a weight to consider the conduction guides, and by default the leads with the higher number of dither events are more conductive to fewer leads (this part writes the algorithm, but is not shown in the figure).

Fig. 9C shows an example of segg electroencephalography data calculation. According to the above method, the 4# lead can be determined as the level 1 node, the 63# lead can be determined as the level 2 node, and the 2# lead can be determined as the level 3 node.

S205: and displaying the drawn high-frequency oscillation conduction network in the individualized brain model.

The individualized brain phantom of the patient may be rendered by Magnetic Resonance Imaging (MRI) techniques. Meanwhile, the positions of the leads on the electrodes are three-dimensionally displayed on the brain model based on preset coordinates of the implanted electrodes, CT scanning images and other data after the electrodes are implanted.

The data about the high-frequency oscillation events detected in step S203 are imported, so that the number, time and energy of the high-frequency oscillations corresponding to each lead can be displayed on the brain model.

Further, the high-frequency oscillation conduction network and its key nodes obtained in step S204 may be mapped and labeled on the brain model accordingly (as exemplarily shown in fig. 12). In this way, the clinician can intuitively understand the distribution of the high frequency oscillations, thereby facilitating the planning and execution of the surgical plan.

S206: and monitoring the dynamic change of the high-frequency oscillation conduction network in real time.

During the operation of an epileptic, partial electrodes can be reserved, and the high-frequency oscillation of the adjacent or far-end position of some suspicious areas and the change condition of the conduction network of the suspicious areas are observed in real time, so that whether the overall epileptic network is sufficiently damaged or not is judged.

When the thermal coagulation damage is carried out, the electroencephalogram signals generated by all the deep electrodes can be monitored in real time because the electrodes do not need to be pulled out. Therefore, the high-frequency oscillation and the change of the conduction network thereof can be detected in real time, and whether the tissue corresponding to the seizure area is completely damaged or not is judged.

The high-frequency oscillation conduction network can reflect the excitability of the whole epileptogenic network, and the obvious weakening or disappearance of the excitability is closely related to better surgical prognosis. The embodiment can be used for the real-time monitoring of various surgeries such as removal of epileptic regions or thermosetting destruction in the surgery, and the like, and the chronic monitoring of the sickroom or the clinic before the surgery, and is suitable for various intracranial recording electrode types commonly used in clinic.

Example 4

In this embodiment, the number of high-frequency oscillation events occurring in each lead is obtained according to the high-frequency oscillation events detected in step S203 in embodiment 3, and the leads in which ripples and fast ripples are detected are arranged in descending order. Leads known to frequently oscillate at high frequencies are often located in epileptogenic regions. Therefore, when an epileptic patient surgically resects an epileptogenic region, the first electrode not to be resected continuously is important in the descending order of the high frequency oscillation leads. When the leads with high rank are sequentially excised according to descending order, if the first uncut lead is positioned in the epilepsy area, the epilepsy area is not completely excised, and epilepsy recurrence is ensured after the operation. Conversely, if this lead is located outside the epileptogenic zone, the subsequent leads will all generate high frequency oscillations at a lower frequency than the lead, so they should not be located inside the epileptogenic zone either. This indicates that the epileptogenic zone has been completely removed and that there should be no recurrence of epilepsy after the procedure. Based on this principle, the ratio between the number of leads from the highest ranked lead to the first unresectable lead and the total number of leads from which high frequency oscillations are detected can be calculated, which is referred to as "high-ranked high frequency oscillation lead continuous ablation ratio".

By comparing this ratio to the patient's prognosis, it was found that the patient can obtain a better prognosis when the high-ranked dither lead continuous ablation ratio is above a threshold (e.g., ≧ 70%). For example, the ablation seizure threshold may be set at 72% of the high-order fast ripple channel. Optionally, after detecting the high-frequency oscillation events, arranging at least 1 high-frequency lead per minute in descending order according to the number of the high-frequency oscillation events, wherein the first 72% of leads are considered to be within the epileptogenic region (rapid ripples are seen preferentially), and the region is defined as a suspicious epileptogenic region to guide surgical resection or heat coagulation destruction treatment.

Example 5

In this example, high frequency oscillations of 26 patients with seizure-causing resection were analyzed retrospectively. 10 patients had no seizures one year after the resection, with good prognosis; the rest of the patients still had either mild or severe seizures, with poor prognosis.

Of these 26 patients, 16 patients used segg electrodes and 10 patients used subdural electrodes to acquire brain electrical signals. Intracranial brain signals were acquired for these patients at 5 minute intervals of onset, high frequency oscillation events in the brain signals were detected, and spurious high frequency oscillation events were removed, as per step 203 in example 2. Finally, a total of 43242 true ripples and 29171 true rapid ripples were detected in 26 patients. At least more than one lead generating high-frequency oscillation events per minute is arranged in descending order according to the number of the generated high-frequency oscillation events, the continuous excision ratio of the high-order high-frequency oscillation leads is calculated and compared with the operation prognosis of each patient.

It was found that the continuous ablation rate of high-ordered hf oscillatory leads was significantly higher in patients with good prognosis than in patients with poor prognosis, whether rippling or rapid rippling (rippling: p 0.018, rapid rippling: p < 0.001). Based on this, the threshold value of the high-ranking high-frequency oscillation lead continuous ablation ratio for assisting to delimit epileptic zone is set at the lower value of 95% confidence interval of continuous ablation high-ranking rapid ripple, namely 72%. That is, when more than 72% of the high-ranked rapid ripple regions are excised, a good prognosis is obtained in at least 95% of patients.

An example of 2 patients is specifically shown in fig. 10, fig. 10A corresponding to patient 1 and fig. 10B corresponding to patient 2. For patient 1, intracranial hf oscillation event detection indicated that ablation should be performed on the left frontal lobe seizure area (as shown in the area marked by the straight line box on the brain phantom in fig. 10A). After a complete resection of this area by surgery (as in the brain model of fig. 10A, the area marked by the curved box), patient 1 had no seizures one year after surgery. For patient 2, intracranial hf oscillation event detection indicated the presence of a large number of hf oscillation events in the left temporal, left parietal border, left frontal area, etc. areas (as indicated by the straight line boxes on the brain model of fig. 10B). However, only the left temporal region was surgically removed (as in the brain model of fig. 10B, the area marked by the curve box), leaving another area that also produced a large number of high frequency oscillation events, resulting in patient 2 still having episodes after surgery.

Example 6

In this example, the high frequency oscillation events of patients with seizure and destruction in epileptic regions were analyzed. Thermocoagulation damage can selectively damage key nodes to interrupt epileptic networks and leave intracranial electrodes all the way.

This example included 26 patients who had suffered a thermal coagulation failure, 10 of whom had no seizures one year after surgery and 16 of whom had seizures.

These 26 patients all used SEEG electrodes to acquire brain electrical signals. Acquiring the electroencephalogram signals of the attack interval of 5 minutes before the damage. Then, according to step S203 in embodiment 3, a high-frequency oscillation event is detected.

According to the step S204 of the embodiment 3, the leads on the SEEG electrode are graded, the high-frequency oscillation event conduction direction is determined, and therefore the high-frequency oscillation conduction network is drawn.

By comparison with the atherogenic destruction and patient prognosis, it was found that patients tended to reach a better prognosis after the destruction level 1 and level 2 nodes. However, considering that the area of heat coagulation damage is extremely limited compared to resection surgery, the final degree of damage and the size of the damaged area need to be further determined due to individual differences. Therefore, the detection of high frequency oscillations after destruction and the analysis of their conduction networks were performed again for these patients. It was found that there was no significant difference between the number of high frequency oscillation events and the intensity of conduction of leads in the conduction network in patients with good and poor prognosis before failure. However, after destruction, the number of remaining events and the conductivity of the leads in the conduction network are significantly reduced in patients with good prognosis (p is less than 0.05).

Furthermore, it has been found that if rapid ripples reappear around the damaged area, a poor prognosis is often predicted. This may suggest that metastasis and abnormal connective remodeling occur in the epileptogenic seizure center.

Fig. 11 shows an example of 1 patient. FIG. 11A shows the

level

1 and 2 nodes determined based on the high frequency oscillatory conduction network, namely leads A3-6 for electrode A and leads E1-4 for electrode E. The leads of the electrodes are ranked according to the method of the present disclosure, and a high-frequency oscillation conduction network is drawn, and the drawn high-frequency oscillation conduction network is fused with the brain model of the patient (exemplarily shown in fig. 12). As can be seen from fig. 11, these level 1 and level 2 nodes are located in the left frontal lobe region and may be closely related to epileptic discharge. Meanwhile, the clinical practical conditions (such as imaging and the like) are combined to carry out thermosetting damage treatment on the leads A1-7 and the leads E1-5 of the left frontal lobe seizure area. The red circles mark some of the electrode points that actually fail. As shown in fig. 11A and 11B, after successful destruction of these electrode sites, the leads that were evident in the pre-destruction discharge were significantly reduced in discharge after destruction, and the high frequency oscillation events detected in real time were also significantly reduced. After comparing the number of high-frequency oscillations before and after the damage with the network, the pathological high-frequency oscillations of the whole brain (especially epileptogenic regions) of the patient after the damage are obviously reduced or even completely disappeared. The upper graph of FIG. 11D shows the variation of the number of high frequency oscillations displayed simultaneously on the patient's brain model before and after the lesion. It can be seen that the high frequency oscillations on the electrodes in the suspected epileptic zone have completely disappeared after the thermal coagulation damage treatment. As shown in fig. 11D, there is a significant high frequency signal conduction before thermal damage, and after the thermal damage process, the high frequency oscillation network is significantly broken or even disappeared. This indicates that the network, which is originally tightly connected around the seizure area, is also destroyed and interrupted after the damage, and is no longer conductive. The patient had no episodes one year after surgery with good prognosis.

In conclusion, the high-frequency oscillation can be monitored in real time during the operation so as to locate the epileptogenic region and carry out prognosis judgment. The high-frequency oscillation is closely related to the epilepsy causing area, and the complete ablation or damage of the high-frequency oscillation area can enable a patient to obtain better prognosis. The excitability of a local epileptogenic network is reflected by the high-frequency oscillation conduction network, and the weakening or disappearance of the high-frequency oscillation conduction network usually indicates better prognosis, so that the monitoring of the dynamic change of the network during the operation has important value for guiding a subsequent treatment scheme or carrying out prognosis judgment.

Example 7

The present embodiment provides the electroencephalogram signal processing apparatus 200 adapted to execute embodiment 2 or embodiment 3. As shown in fig. 13, the electroencephalogram signal processing apparatus 300 includes an acquisition module 301, a processing module 302, a first calculation module 303, and a second calculation module 304.

The acquisition module 301 may be used to acquire brain electrical signals of a subject. The processing module 302 may be configured to select one or more time segments of the electroencephalogram signal from the acquired electroencephalogram signals. The first calculation module 303 may be configured to obtain a high frequency oscillation event in the brain electrical signal of the one or more time periods. A second calculation module 304 is operable to plot a high frequency oscillation conduction network based on the detected high frequency oscillation events, wherein the second calculation module is operable to randomly pair the leads, each pair of leads comprising a first lead and a second lead, calculate a ratio R according to the following formula,

The first calculation module 303 may further comprise a traversal module 403 for determining parameters for detecting the high-frequency oscillation event, wherein the parameters comprise a low threshold, a high threshold and the number of high-frequency oscillation peak points satisfying the high threshold.

The first calculating module 303 may further include a baseline determining module 404, configured to arrange the absolute amplitudes of the high-frequency oscillatory waves in the electroencephalogram signal from small to large to obtain a maximum peak point distribution curve, determine a turning point on the maximum peak point distribution curve, and use an average value of the absolute amplitudes of all oscillatory waves before the turning point as a baseline. The baseline determination module 404 may be configured to determine a dynamic baseline of the brain electrical signal for the one or more time periods using a moving time window over the one or more time periods.

The second calculation module 304 further includes a decision module 405 for ranking the leads based on the magnitude of the ratio R.

The brain electrical signal processing apparatus 300 may further comprise a fusion module for fusing the high-frequency oscillatory conduction network to an individualized three-dimensional brain model of the subject. Three-dimensional brain models can be obtained by magnetic resonance imaging techniques.

The brain electrical signal processing apparatus 300 may further include a monitoring module for monitoring the number of high frequency oscillation events and/or changes in the high frequency oscillation conduction network.

The technical solution of the present disclosure is not limited to the limitations of the above specific embodiments, and all technical modifications made according to the technical solution of the present disclosure fall within the protection scope of the present disclosure.

Claims

1. A method of mapping a high frequency oscillatory conduction network comprising the steps of:

acquiring high-frequency oscillation events in the electroencephalogram signals of one or more time periods; and

drawing a high-frequency oscillation conducting network based on the high-frequency oscillation event, wherein the drawing the high-frequency oscillation conducting network comprises: randomly pairing all leads, wherein each paired lead set comprises a first lead and a second lead, calculating a ratio R according to the following formula,

2. A method for brain electrical signal processing, comprising the steps of:

collecting electroencephalogram signals of a subject;

selecting one or more time-period electroencephalogram signals from the acquired electroencephalogram signals;

acquiring high-frequency oscillation events in the electroencephalogram signals of the one or more time periods; and

drawing a high-frequency oscillation conducting network based on the high-frequency oscillation event, wherein the drawing the high-frequency oscillation conducting network comprises: randomly pairing leads, wherein each paired lead set comprises a first lead and a second lead, calculating a ratio R according to the following formula,

3. The method according to claim 1 or 2, wherein the step of obtaining the high-frequency oscillation events in the electroencephalogram signals of one or more time periods comprises the steps of arranging the absolute amplitudes of the high-frequency oscillation waves in the electroencephalogram signals from small to large to obtain a maximum peak point distribution curve, determining a turning point on the maximum peak point distribution curve, and taking the average value of the absolute amplitudes of all oscillation waves before the turning point as a baseline;

preferably, a dynamic baseline of the brain electrical signal for the one or more time periods is determined using a moving time window over the one or more time periods.

4. The method of claim 3, wherein the step of acquiring high frequency oscillation events in the brain electrical signal for one or more time periods further comprises:

when a grid-shaped or sheet-shaped electrode is used for collecting electroencephalogram signals, after 80-200 Hz filtering is carried out, when 8 or more continuous wave peak point vibration amplitudes are higher than 3.5 times of standard deviation of the base line and 6 or more continuous wave peak point vibration amplitudes are higher than 9 times SD of the base line, the high-frequency oscillation event is judged to be ripple; and/or after 200-500 Hz filtering, when the amplitude of continuous 8 or more peak points is higher than 3 times of standard deviation of the baseline and 4 or more peak points are higher than 10.5 times of standard deviation of the baseline, judging that the high-frequency oscillation event is rapid ripple; or

When a stereo directional electrode is used for collecting electroencephalogram signals, after 80-200 Hz filtering is carried out, when 8 or more continuous wave peak point amplitudes are higher than 3 times of standard deviation of the base line, and 6 or more wave peak points are higher than 10 times of standard deviation of the base line, the high-frequency oscillation event is judged to be ripple; and/or after 200-500 Hz filtering, judging that the high-frequency oscillation event is a rapid ripple when continuous 8 or more peak point amplitudes are higher than 3 times of standard deviation of the base line and 6 or more peak point amplitudes are higher than 9.5 times of standard deviation of the base line.

5. The electroencephalogram signal processing method according to any one of claims 1 to 4, wherein the step of obtaining high-frequency oscillation events in the electroencephalogram signals of one or more time periods further comprises removing false high-frequency oscillation events, wherein frequency offset energy differences are detected based on a spectral density map drawn by a spectral phase space reconstruction method, the frequency offset energy differences of the frequency bands where the high-frequency oscillation events are located are accumulated, if the obtained value is less than or equal to 0, the corresponding high-frequency oscillation event is judged to be a false high-frequency oscillation event and removed,

preferably, the false high frequency oscillation event is caused by the gibbs effect.

6. The method according to any of claims 1 to 5, wherein the high frequency oscillation event occurring on the second lead associated with the first lead is a high frequency oscillation event occurring on the second lead within a time period t after the high frequency oscillation event occurred on the first lead, preferably a high frequency oscillation event conducted from the first lead to the second lead,

preferably, the time period t is 100-300 ms.

7. The method according to any one of claims 1 to 6, wherein said plotting said high-frequency oscillating conducting network based on said ratio R comprises: ranking the leads based on the magnitude of the ratio R,

in all paired lead groups with one lead as a first lead, when more than 2 ratios R in all calculated ratios R are more than or equal to 0.5, judging that the lead actively conducts a high-frequency oscillation event; and/or

In all paired lead groups with one lead as the second lead, when the ratio R of less than 2 in all calculated ratios R is more than or equal to 0.5, the lead is judged not to receive the high-frequency oscillation event.

8. The method according to any one of claims 1 to 7, wherein said plotting said high-frequency oscillating conducting network based on said ratio R comprises:

when one lead actively conducts a high-frequency oscillation event and does not receive the high-frequency oscillation event, judging the lead as a level 1 node;

when the first lead in the paired lead group is judged to be the level 1 node, if the ratio R of the paired lead group is more than or equal to 0.5 and the second lead actively conducts the high-frequency oscillation event, the second lead of the paired lead group is judged to be the level 2 node, otherwise the second lead is judged to be the level 3 node;

when the first lead in the paired lead group is judged to be the level 2 node, if the ratio R of the paired lead group is more than or equal to 0.5, the second lead of the paired lead group is judged to be the level 3 node;

preferably, in the high frequency oscillation conduction network, the high frequency oscillation event is conducted from the level 1 node to the level 2 node and from the level 2 node to the level 3 node.

9. An apparatus for mapping a high frequency oscillatory conduction network, comprising:

the first calculation module is used for acquiring high-frequency oscillation events in the electroencephalogram signals of the one or more time periods;

a second calculation module for mapping a high frequency oscillation conduction network based on the high frequency oscillation events, wherein the second calculation module is used for randomly pairing leads, each paired lead group comprises a first lead and a second lead, and the ratio R is calculated according to the following formula,

10. An apparatus for brain electrical signal processing, comprising:

the acquisition module is used for acquiring electroencephalogram signals of a subject;

the processing module is used for selecting one or more time-period electroencephalogram signals from the acquired electroencephalogram signals;

11. The apparatus according to claim 9 or 10, wherein the first calculation module further comprises a baseline determination module, the baseline determination module is configured to arrange the absolute amplitudes of the high-frequency oscillatory waves in the electroencephalogram signal from small to large to obtain a maximum peak point distribution curve, determine a turning point on the maximum peak point distribution curve, and use the average value of the absolute amplitudes of all oscillatory waves before the turning point as a baseline;

preferably, the baseline determination module is configured to determine the dynamic baseline of the brain electrical signal for the one or more time periods using a moving time window over the one or more time periods.

12. The apparatus of any of claims 9 to 11, wherein the first computing module further comprises a traversal module configured to determine parameters for detecting the high-frequency oscillation event, wherein the parameters include a low threshold, a high threshold, and a number of high-frequency oscillation peak points that satisfy the high threshold.

13. The apparatus according to any one of claims 9 to 12, wherein the high frequency oscillation event occurring on the second lead associated with the first lead means that a high frequency oscillation event also occurs on the second lead within a time period t after the high frequency oscillation event occurs on the first lead, preferably a high frequency oscillation event conducted from the first lead to the second lead,

preferably, the time period t is 100-300 ms.

14. The apparatus according to any one of claims 9 to 13, wherein the second calculation module further comprises a decision module for ranking the leads based on the magnitude of the ratio R;

preferably, the determination module is configured to:

in all paired lead groups with one lead as a first lead, when the calculated ratio R of 2 or more than 2 in all the ratios R is more than or equal to 0.5, judging that the lead actively conducts a high-frequency oscillation event; and/or

In all paired lead groups with one lead as the second lead, when the calculated ratio R of 2 or less than 2 in all the ratios R is more than or equal to 0.5, the lead is judged not to receive the high-frequency oscillation event.

15. The apparatus of claim 14, wherein the determination module is configured to:

16. The apparatus according to any one of claims 9 to 15, further comprising a fusion module for fusing the high frequency oscillation conduction network to a three-dimensional brain model of the subject, preferably obtained by means of magnetic resonance imaging techniques,

more preferably, the apparatus further comprises a monitoring module for monitoring the number of high frequency oscillation events and/or changes in the high frequency oscillation conduction network.

17. A computer-readable storage medium having a program stored thereon, wherein the program, when executed by a processor, implements the method of any of claims 1-8.

18. A surgical system comprising the apparatus of any one of claims 9 to 16 and/or the computer-readable storage medium of claim 17, and/or performing the method of any one of claims 1 to 8.

19. A model of a high frequency oscillatory conduction network, when plotted by the method of any one of claims 1 to 8.