Method and device for detecting epileptic seizures

Abstract

The present application proposes a method and a device for detecting epileptic seizures. The method includes obtaining electroencephalogram signals of a subject on multiple lead channels, where the electroencephalogram signals have multiple signal segments; extracting candidate signal segments from the multiple signal segments based on the temporal characteristics and frequency characteristics of the electroencephalogram signals; and analyzing the correlation index of the candidate signal segments to determine whether the candidate signal segments indicate epileptic seizures. The solution of the present application can obtain a more accurate and sensitive detection result of epileptic seizures.

Description

Background Art

[0002] Epilepsy, one of the common diseases of the human nervous system, is commonly known as "hysterical fits", which is characterized by repeated epileptic seizures caused by abnormal discharge of brain neurons. According to statistics, there are approximately 200 to 300 epilepsy patients per 100,000 people. Since epileptic patients may experience whole-body spasms and are likely to lose consciousness when they suddenly have an attack, various risks are hidden in the epilepsy disorder.

[0003] Currently, in addition to clinical observation, the diagnosis and detection of epileptic seizures in epileptic patients are mainly achieved through the analysis of electroencephalogram (EEG) signals. The analysis and diagnosis of long-term EEG signals are mainly completed by doctors through manual reading of graphs and experience. This manual detection method is time-consuming and laborious, cannot support quantitative analysis, resulting in varying accuracy from person to person, and cannot provide a unified and effective detection standard.

[0004] EEG is a signal data form commonly used to record the electrical activities generated by the human cerebral cortex. The characteristics of high temporal resolution, non-invasiveness, and low cost of EEG signals make them widely used in the evaluation of patients with epileptic foci. Using computer-aided epilepsy seizure detection of EEG signals can reduce the workload of doctors, improve the efficiency, accuracy, and consistency of detection, and reduce the diagnosis and treatment costs of epilepsy detection. However, the existing automated detection technologies do not distinguish the signal waveform characteristics between the EEG signals of epileptic seizure discharges and the EEG signals with artifacts, resulting in a relatively high false detection rate. In addition, the existing detection schemes usually perform epilepsy seizure detection on the EEG signals during specific types of epileptic seizures, without comprehensively considering the characteristics of different types of epileptic seizures, resulting in relatively low sensitivity of epilepsy detection and poor detection effects.

[0005] Therefore, there is a need to improve the epilepsy seizure detection scheme for EEG signals. Summary of the Invention

[0006] In view of the defects and problems to be solved mentioned above, the present application proposes a method, device, and computer storage medium for detecting epileptic seizures, which comprehensively consider the characteristics of different epileptic seizure types in terms of signal frequency and duration length, so as to effectively improve the accuracy and sensitivity of detection results.

[0007] According to one aspect of the present application, a method for detecting epileptic seizures is proposed, including: obtaining electroencephalogram signals of a subject on multiple lead channels, where the electroencephalogram signals have multiple signal segments; extracting candidate signal segments from the multiple signal segments based on the time characteristics and frequency characteristics of the electroencephalogram signals; analyzing the correlation index of the candidate signal segments to determine whether the candidate signal segments indicate epileptic seizures.

[0008] According to another aspect of the present application, there is provided a device for detecting epileptic seizures, including: an acquisition unit configured to acquire electroencephalogram (EEG) signals of a subject on multiple lead channels, the EEG signals having multiple signal segments; and a detection unit configured to extract candidate signal segments from the multiple signal segments based on the temporal characteristics and frequency characteristics of the EEG signals, and analyze the correlation index of the candidate signal segments to determine whether the candidate signal segments indicate epileptic seizures.

[0009] According to still another aspect of the present application, there is provided a computer-readable storage medium having stored thereon a computer program, the computer program including executable instructions that, when executed by a processor, implement the method as described above.

[0010] According to yet another aspect of the present application, there is provided an electronic device including a processor and a memory for storing executable instructions of the processor; wherein the executable instructions, when executed by the processor, implement the method as described above.

[0011] By adopting the solution for detecting epileptic seizures proposed in the embodiments of the present application, suspicious candidate signal segments are extracted respectively based on the frequency characteristics and temporal characteristics of EEG signals, where the characteristics of prominent low-frequency and high-frequency signal energies in different types of epileptic seizures are emphasized, and the duration lengths of different types of epileptic seizure conditions are specifically analyzed. Further, the correlation index of the EEG signals on multiple lead channels is analyzed during the extraction of candidate signal segments to screen out accurate signal segments indicating epileptic seizures. This solution can identify epileptic seizures from EEG signals of multiple information channels with a more refined detection criterion. In particular, it can effectively distinguish the signal characteristics between the EEG signals during epileptic seizures and the EEG signals with artifacts. It can be applied to any EEG signal containing abnormal discharge waveforms during epileptic seizures, and the signal characteristics of signal segments are examined based on the duration of different types of epileptic seizure discharges without continuously monitoring the EEG signals for a long time, thereby obtaining detection results with higher accuracy and sensitivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] By referring to the accompanying drawings and describing its exemplary embodiments in detail, the above and other features and advantages of the present application will become more apparent.

[0013] Figure 1 It is a schematic flowchart of a method for detecting epileptic seizures according to an embodiment of the present application.

[0014] Figure 2A It is an exemplary low-frequency EEG waveform diagram during an epileptic seizure according to an embodiment of the present application.

[0015] Figure 2BExemplary high-frequency EEG waveform diagram during a seizure according to an embodiment of the present application.

[0016] Figure 3A Exemplary EEG waveform diagram with a long candidate signal segment according to an embodiment of the present application.

[0017] Figure 3B Exemplary EEG waveform diagram with a short candidate signal segment according to an embodiment of the present application.

[0018] Figure 4 Exemplary EEG waveform diagram of a candidate signal segment with an artifact signal according to an embodiment of the present application.

[0019] Figure 5A Exemplary EEG waveform diagram of a candidate signal segment for a correlation index to be analyzed according to an embodiment of the present application.

[0020] Figure 5B Exemplary EEG waveform diagram of another candidate signal segment for a correlation index to be analyzed according to an embodiment of the present application.

[0021] Figure 6 Schematic structural block diagram of a device for detecting seizures according to an embodiment of the present application.

[0022] Figure 7 Schematic block diagram of an electronic device for detecting seizures according to an embodiment of the present application. Detailed implementation

[0023] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, the exemplary embodiments can be implemented in various forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that the present application will be thorough and complete, and the concept of the exemplary embodiments will be fully conveyed to those skilled in the art. In the figures, for clarity, the dimensions of some elements may be exaggerated or deformed. Identical reference numerals in the figures denote identical or similar structures, and thus their detailed descriptions will be omitted.

[0024] In addition, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to give a thorough understanding of the embodiments of the present application. However, those skilled in the art will realize that the technical solutions of the present application can be practiced without one or more of the specific details, or other methods, elements, etc. can be adopted. In other cases, well-known structures, methods, or operations are not shown or described in detail to avoid obscuring various aspects of the present application.

[0025] Clinically, an electroencephalogram (EEG) device (such as an EEG amplifier) is used to collect the EEG signals of patients. Seizure discharges related to epileptic foci can be detected through various spike waveforms, various sharp wave waveforms, waveforms of complex slow waves such as spike-slow waves and sharp-slow waves, and various waveforms of low-amplitude fast activities in the EEG signals. In the actual reading and diagnosis of EEG signals, these waveforms related to epileptiform discharges account for most cases of epileptiform discharges.

[0026] The durations of epileptic symptoms of different seizure types are also different. Reflected in the EEG signal waveforms, the EEG signals collected during different types of seizures also have corresponding characteristics in terms of frequency, duration, signal energy, and waveform morphology. The duration (abbreviated as duration) of most seizure types during the seizure period ranges from several seconds to dozens of seconds. Among them, the durations of myoclonic seizures and atonic seizures range from dozens to hundreds of milliseconds, and are accompanied by prominent electromyogram signal characteristics. For example, in terms of the signal energy related to frequency in the EEG signals collected during seizures, some seizure types show prominent high-frequency energy during seizures while some seizure types show prominent low-frequency energy during seizures.

[0027] Therefore, seizure types can be classified based on the corresponding EEG signal characteristics during seizures. For example, time characteristics and frequency characteristics can relatively accurately distinguish the types of seizure events in different situations, thereby helping to extract the EEG signal segments during seizures from the EEG signals. According to the seizure duration and frequency band range, it can be divided into four different situations: signal segments with long duration (also called long-duration high-frequency) with high-frequency characteristics, signal segments with long duration (long-duration low-frequency) with low-frequency characteristics, signal segments with short duration (short-duration high-frequency) with high-frequency characteristics, and signal segments with short duration (short-duration low-frequency) with low-frequency characteristics. Among them, myoclonic seizures and atonic seizure types belong to short-time signal segments, and seizure types of long-time signal segments include tonic seizures, typical absence seizures, atypical absence seizures, and focal seizures, etc. According to the embodiments of the present application, the signal segments are first divided into long-time type and short-time type signal segments according to the duration, and then for different duration types, the characteristics of the signal segments in different frequency band ranges are further extracted for analysis. That is, among various seizure types of signal segments with durations belonging to short-time or long-time respectively, they are further subdivided according to their energy distributions in different frequency band ranges (such as high frequency or low frequency).

[0028] The continuously acquired electroencephalogram (EEG) signals can be divided into multiple signal segments. The EEG signals can be divided into different signal segments based on the correspondence between their signal time characteristics and frequency characteristics and the corresponding seizure types. These signal segments can have the same or different durations (e.g., long-duration signal segments and short-duration signal segments). The EEG signals include multiple lead channels, and for each lead channel, the signal segments that may have seizures can be extracted from the EEG signals according to the time characteristics and frequency characteristics. These extracted signal segments are suspicious signal segments whose seizure indication is undetermined and are referred to as candidate signal segments in this article.

[0029] In the selection of frequency characteristics or frequency bands, high-frequency and low-frequency characteristics are mainly considered because in common seizure types, the usual EEG signal waveforms include at least one of low-amplitude fast activities, polyspike waveforms, sharp wave waveforms, and complex waveforms of spike and slow waves in the high-frequency and low-frequency bands, and there are few seizure types with prominent signal energy only in the mid-frequency band range. Therefore, considering only the frequency characteristics of the EEG signals in the mid-frequency band usually does not bring more accurate detection effects.

[0030] The following combines Figure 1 the schematic flow of the system and method for detecting seizures shown in Figures 2A to 6 to introduce the technical solution of this application in detail. Among them, the example EEG signal waveforms detected in different situations and steps are shown in

[0031] The seizure detection method according to the embodiment of this application at least includes step S110 of acquiring EEG signals of multiple lead channels, step S130 of extracting candidate signal segments based on the time characteristics and frequency characteristics of the EEG signals, and step S140 of analyzing the correlation index of the candidate signal segments to determine whether the candidate signal segments indicate seizures. Optionally, before acquiring the EEG signals and before extracting the candidate signal segments, step S120 of preprocessing the EEG signals can be included, which is shown in Figure 1 with a dashed box.

[0032] First, in step S110, the method acquires EEG signals of multiple lead channels from an electroencephalogram device. The electroencephalogram (EEG) signals can be collected using an electroencephalogram amplifier according to clinical standards, and its sampling rate can be above 200 Hz and the number of lead channels can be above 16. In Figures 2A to 6 the corresponding waveforms of the EEG signals are shown with multiple lead channels. These lead channels include, for example, FP1, FP2, F3, F4, C3, C4, P4, O1, F7, T3, FZ, CZ, and PZ, etc. The EEG signals of multiple lead channels are synchronously recorded on the same EEG signal waveform diagram on the time (in seconds) horizontal axis for comparison.

[0033] In the optional step S120, preprocessing operations are performed on the EEG signals. The purpose of the preprocessing operations is to reduce the amount of data calculation to improve the calculation efficiency, eliminate the influence of invalid data and interference on the detection, and improve the resolution and accuracy of the EEG signal waveform, etc.

[0034] The preprocessing operation step S120 may include at least one of the sub-step S111 of resampling the EEG signal, the sub-step S112 of removing the EEG signals on the invalid lead channels, the sub-step S113 of filtering to remove power frequency noise, baseline drift noise, and / or high-frequency noise, and the sub-step S114 of calibrating the EEG signal (re-referencing) based on a reference signal.

[0035] The sub-step S111 of resampling the EEG signal is used to improve the calculation efficiency. The resampling frequency should be guaranteed to be more than twice the sampling frequency of the EEG amplifier, generally above 150 Hz, for example, selected as 200 Hz as described above, so as to ensure the detection accuracy. Resampling can be achieved by downsampling or by upsampling through interpolation.

[0036] The purpose of the sub-step S112 of removing the EEG signals on the invalid lead channels is to screen out the signals of the lead channels that are no longer needed in the later steps, such as the EEG signals of the lead channels of the bilateral mastoid points used as references. The criterion for judging whether the EEG signal of a lead channel is needed can be whether the lead channel effectively collects the EEG signal.

[0037] The sub-step S113 of filtering to remove noise is used to avoid the negative impact of various interference signals on the detection. These noises include power frequency noise, baseline drift noise, and / or high-frequency noise, etc. Power frequency noise is generally the noise caused by the inherent frequency of the power system. In our country, the working frequency of alternating current is 50 Hz (or 60 Hz). Power frequency noise will interfere with the normal operation of electronic devices and may cover the effective information related to the detection of epileptic seizures contained in the electroencephalogram signal. A notch filter with a working frequency of 50 Hz and / or multiples of 50 Hz can be designed to eliminate the power frequency noise interference in the EEG signal. Filters can also be designed to remove the baseline drift noise and / or high-frequency noise in the signal. Baseline drift mainly comes from the change of the parameters of circuit components caused by the physical characteristics of the channel sensor and environmental factors (such as environmental temperature changes), which results in the superposition of components unrelated to the measured signal in the signal input. High-frequency noise comes from electromyogram interference and environmental noise. A band-pass filter with a working frequency range of 1.6 - 70 Hz can be designed to filter and denoise the EEG signal. The working frequency range can be set according to experience. In principle, the frequency band characteristics of the EEG signal during epileptic seizures are mainly concentrated in the frequency range of 1.6 - 70 Hz.

[0038] The sub-step S114 of rereferencing is used to calibrate the EEG signal based on the reference signal, where the EEG signal detected on multiple global lead channels is subtracted from the reference signal to eliminate the common error existing in the signals on all lead channels. For example, there may be a seizure in the EEG signal on the lead channel of the left forehead, while there is no seizure in the EEG signal obtained on the lead channel of the right forehead. However, there is a common error signal in both of them. Rereferencing can remove the common error signal in both of them. Rereferencing can include rereferencing algorithms such as monopolar rereference, bipolar rereference, laplacian rereference, or common average rereference. In this article, the common average rereference is used as an example to introduce the method of this application, that is, the average value of all EEG signal data of the whole brain is used as the reference data. The data obtained by subtracting the reference signal value at each sampling moment from the waveform amplitude value of the acquired EEG signal at each sampling moment can obtain the EEG signal after rereferencing.

[0039] After the preprocessing of the EEG signal is completed, in step S130, candidate signal segments that may indicate or in which there is a seizure are extracted. As described above, four typical candidate signal segments of long-term high-frequency, long-term low-frequency, short-term high-frequency, and short-term low-frequency classified according to time characteristics and frequency characteristics may indicate the presence of a seizure in them. Step S130 can initially screen the EEG signal segments in the high-frequency signal components and / or low-frequency signal components of the EEG signal that protrude from the background signal of the EEG signal. The extracted candidate signal segments can be merged with adjacent candidate signal segments according to the duration length of a real seizure, so as to provide for the next step to analyze in detail whether there is a real seizure or the signal segments that are truly in the seizure period in the candidate signal segments. Among them, the background signal is a signal segment with a low signal energy value in the EEG signal after high-frequency or low-frequency filtering.

[0040] Step S130 may include sub-step S131 of extracting the high-frequency EEG signal and the low-frequency EEG signal of the EEG signal respectively for long-term type and short-term type signal segments, sub-step S132 of screening the signal segments respectively for the extracted high-frequency EEG signal and low-frequency EEG signal, and sub-step S133 of merging the screened signal segments and extracting those candidate signal segments that exceed the time range threshold.

[0041] According to the embodiments of the present application, these selected EEG signal segments are further analyzed for low-frequency and high-frequency EEG signals according to long-time signal segment types and short-time signal segment types respectively. For example, in the long-time signal segment type, the extracted high-frequency EEG signals are used to analyze the candidate signal segments of long-time high-frequency, and the extracted low-frequency EEG signals are used to analyze the candidate signal segments of long-time low-frequency. Similarly, in the short-time signal segment type, the extracted high-frequency EEG signals are used to analyze the candidate signal segments of short-time high-frequency, and the extracted low-frequency EEG signals are used to analyze the candidate signal segments of short-time low-frequency.

[0042] Therefore, in sub-step S131, for the signal segments of long-time and short-time types respectively, the EEG signals are filtered in the low-frequency band range of 1.6 - 5 Hz and the high-frequency band range of 14 - 40 Hz. Among multiple signal segments, the long-time type signal segments and short-time type signal segments corresponding to epileptic seizures of different duration lengths are extracted, and then for the long-time type and short-time type signal segments respectively, high-frequency electroencephalogram signals and low-frequency electroencephalogram signals are extracted for further analysis. According to statistical and empirical data, the waveform characteristics of EEG signals during epileptic seizures are mainly reflected in the low-frequency band range of 1.6 - 5 Hz and the high-frequency band range of 14 - 40 Hz.

[0043] Figure 2A Shows the low-pass filtered EEG signals, where there are large-amplitude low-frequency spikes that appear on multiple lead channels at position 201a during the period from the 4191st second to the 4194th second and at position 202a near the 4194th second, there is a low-frequency spike and slow wave complex with relatively small waveform amplitude at position 203a near the 4200th second, and there are low-frequency sharp waves on a few lead channels (such as FP1, FP2, F3, F7, and F8) at position 204a after the 4203rd second. These EEG signal segments with prominent energy may all have epileptic seizures. Figure 2B Then shows Figure 2A the waveform of the EEG signal in Figure 2A after high-pass filtering, where there are high-frequency spikes with larger waveform amplitudes than those in the high-frequency EEG signal components that appear on multiple lead channels at position 201b during the period from the 4191st second to the 4194th second (corresponding to 201a of Figure 2A ) and at position 202b near the 4194th second (corresponding to 202a), there are no high-frequency signal segments with significantly prominent energy at the position corresponding to 203a of Figure 2A near the 4200th second, while there are high-frequency sharp waves on the same few lead channels at position 204b (corresponding to 204a) after the 4203rd second. Based only on 2BFrom the signal waveforms, it can be seen that some EEG signal segments with prominent energy are simultaneously reflected in the high-frequency and low-frequency signal components at the same time. The possibility of true epileptic seizures in these signal segments is higher than that of those signal segments with prominent energy only in one of the high-frequency and low-frequency signal components.

[0044] It can be seen that Figure 2A and 2B present the EEG signal waveforms during epileptic seizures, while the EEG signal waveforms during the interictal period when the subject is in the vast majority of time have no obvious difference from the normal EEG signal waveforms of brain waves. However, abnormal discharges also occur at the brain lesions of the subject during the interictal period, and waveforms such as sharp waves, spike waves, spike-slow waves, and sharp-slow waves will also appear in the EEG signals collected at this time. The discharges during epileptic seizures can be considered as the continuity and diffusion of the EEG signal waveforms during the interictal period. For example, if widespread 3Hz spike-slow complex waves appear sporadically or last for less than 2 - 3 seconds, it will not cause absence seizures, and this EEG signal segment belongs to interictal discharges. However, if the duration of this spike-slow complex wave exceeds 5 seconds, the impairment of the subject's consciousness can be observed, and this EEG signal segment belongs to the discharges during epileptic seizures.

[0045] Therefore, in sub-step S132, it is necessary to screen out those signal segments that meet the epileptic seizure discharge phenomenon as suspicious candidate signal segments by comparing the waveform amplitudes of the high-frequency EEG signal and the low-frequency EEG signal with the corresponding waveform amplitude thresholds. The waveform amplitude thresholds for EEG signals in different frequency bands can be set according to the statistical results of the existing EEG signal data during epileptic seizures, and threshold detection is performed on the signal segments to be detected to extract the signal segments that exceed the waveform amplitude threshold. For example, the existing statistical results of the EEG signal during epileptic seizures can be generated based on the data of those signal segments diagnosed as having epileptic seizures in the long-term EEG signal data of different epileptic patients.

[0046] Since the signal segments are further analyzed according to the frequency characteristics on the basis of classifying the signal segments according to the time characteristics of different types of epileptic seizures, it is necessary to set the EEG signal waveform amplitude thresholds corresponding to different signal duration lengths for the high-frequency and low-frequency EEG signals respectively in different signal segment types with different signal duration lengths. For example, a long-term high-frequency waveform amplitude threshold and a long-term low-frequency waveform amplitude threshold can be set for the long-term type of signal segments, and a short-term high-frequency waveform amplitude threshold and a short-term low-frequency waveform amplitude threshold can be set for the short-term type of signal segments.

[0047] After filtering out signal segments that may have epileptic seizures in sub-step S132, an operation of combining the filtered signal segments can also be performed in an optional sub-step S133. The combining operation can combine the filtered signal segments that are adjacent in time into an overall combined signal segment, and select the combined signal segments within a certain duration range as candidate signal segments. The signal segments filtered out in sub-step S132 may be multiple time intervals that are adjacent or non-adjacent in time. Here, the combination of adjacent long-term or short-term signal segments after filtering is for combining the time intervals on the horizontal axis rather than the combination of EEG signals on different lead channels. For example, the EEG signals in the two time intervals from 3 seconds to 5 seconds and from 5.5 seconds to 7 seconds are combined into a synthetic EEG signal with a time interval from 3 seconds to 7 seconds. Therefore, the coordinates of the combined EEG signal or signal segment on the time horizontal axis are still sequential. The combined EEG signal is still a combination of EEG signal segments on multiple lead channels. The purpose of the combining operation is to combine adjacent time windows into an overall time window, and the combined EEG signal segments in this overall time window may include a single epileptic seizure. A single epileptic seizure can be a single long-term epileptic seizure or a single short-term epileptic seizure. Here, the EEG signal corresponding to the combined overall time window may include one or more of the filtered signal segments. The combined filtered signal segments should be compared with the time range conditions corresponding to long-term epileptic seizures and short-term epileptic seizures respectively. Only those combinations of the combined filtered signal segments that meet the corresponding time range conditions are considered to be related to long-term epileptic seizures or short-term epileptic operations, and are called combined signal segments. The time range conditions can be that the duration of the time window or the combination of the filtered signal segments is within the predetermined (long-term or short-term) epileptic seizure duration range, or coincides with or deviates within a set threshold range from the predetermined epileptic seizure duration range. Figure 3A An EEG signal waveform showing a long-term candidate signal segment 301 is presented, where the combined filtered signal segments in the candidate signal segment correspond to an overall time window from 22 seconds to 54 seconds, and the 32-second duration meets the duration range of a specific type of long-term epileptic seizure. It can be seen that the EEG signal shows a dense spike waveform during this period. Figure 3B Then there is an EEG signal waveform with a short-term candidate signal segment 302, where the combined filtered signal segments in the candidate signal segment correspond to an overall time window from 8 seconds to 10 seconds, and the 2-second duration meets the duration range of another specific type of short-term epileptic seizure. Similarly, in Figure 3B it can be seen a Figure 3A similar dense sharp wave waveform.

[0048] After step S130, the signal segments of the EEG waveforms that may have seizure discharges after preliminary selection are further analyzed in step S140. Step S130 extracts or preliminarily screens suspicious signal segments or combinations of signal segments from the perspective of the time characteristics and frequency characteristics of the signal, while step S140 further screens the candidate signal segments that truly exist or indicate seizures by judging the signal correlation between the lead channels in the multi-lead channel EEG signal.

[0049] According to an embodiment of the present application, the analysis in step S140 can be based on the full-band EEG signals on each lead channel. Since the candidate signal segments of the long-time signal segment type and the short-time signal segment type have different various thresholds and correlation thresholds respectively, the corresponding high-frequency EEG signals and low-frequency EEG signals of the candidate signal segments of the long-time signal segment type and the short-time signal segment type extracted in step S130 can be analyzed respectively.

[0050] The correlation index can include energy correlation and waveform correlation. Step S140 of analyzing the correlation index of the candidate signal segments to determine whether there is a real seizure at least includes a sub-step S143 of calculating the energy correlation value between the EEG signals in the candidate signal segments on different lead channels, a sub-step S144 of calculating the waveform correlation value between the EEG signals in the candidate signal segments on different lead channels, and a sub-step S145 of comparing the calculated energy correlation value and waveform correlation value with the corresponding energy correlation threshold and waveform correlation threshold respectively to determine whether there is a seizure.

[0051] According to an embodiment of the present application, before calculating the energy correlation value and the waveform correlation value, there may be a sub-step S141 of dividing the candidate signal segment into time windows and a sub-step S142 of identifying and screening out the ineffective lead channels for the divided time windows, as Figure 1 shown by the dashed box in.

[0052] In the dividing sub-step S141, the candidate signal segment is divided into time windows with a length of 3 seconds (hereinafter referred to as candidate signal sub-segments). If the duration length of the candidate signal segment is less than 3 seconds, there is no need to further divide the candidate signal segment. The time length as the time window dividing standard is determined according to the synchronization time of the EEG signal waveform during the seizure period. According to experience and statistics, the synchronization of the EEG signal waveform during the seizure period is very high within 3 seconds in the vast majority of cases. Therefore, using a 3-second length as a criterion to judge whether the EEG signals on the lead channels within the time window are ineffective has a high credibility.

[0053] Within the segmented time window, it is determined whether the absolute value of the waveform amplitude of the EEG signal of the candidate signal sub - segment on the lead channel meets the condition for determining the failure of the lead channel. The condition for determining the failure of the lead channel can be that when the maximum value of the absolute value of the waveform amplitude of the EEG signal of the candidate signal sub - segment exceeds a preset waveform amplitude threshold, it is determined that the lead channel where the candidate signal sub - segment is located fails, so that the candidate signal sub - segment (time window) on this lead channel is excluded from the calculation of the correlation index. The purpose of determining whether the lead channel fails is to exclude the artifact signal (artifact) in the EEG signal. The artifact signal is generated, for example, due to the limb movement of the subject or the interference of the surrounding environment. The waveform amplitude threshold can be set to 400 μV, for example, so that when the maximum value of the absolute value of the waveform amplitude of the EEG signal of the candidate signal sub - segment on the lead channel exceeds 400 μV, this lead channel is considered to be affected by the artifact signal and has no detection value. As Figure 4 shown, for the candidate signal segment 401, at least in two 3 - second time windows (candidate signal sub - segments) from the 0th second to the 3rd second and from the 9th second to the 12th second, the maximum values of the absolute values of the waveform amplitudes of the EEG signals on the lead channels F4, F8, and T4 all exceed 400 μV. Therefore, these lead channels are determined to be the failed lead channels with artifact signals in the EEG signal. Figure 5A And Figure 5B respectively show the candidate signal segments 501 and 502 without artifact signals, and there is no case where the maximum value of the absolute value of the waveform amplitude of their EEG signals exceeds the waveform amplitude threshold in each 3 - second - long candidate signal sub - segment on all conduction channels.

[0054] According to the setting, if the EEG signals on more than three lead channels are determined to be failed within the candidate signal sub - segment (time window), it is considered that the candidate signal sub - segment or the time window is invalid, thus effectively reducing the calculation burden of the correlation index. Usually, the number of lead channels of the acquired EEG signal exceeds 16, so there will be no situation where only the EEG signal segments on less than three lead channels are extracted as suspicious candidate signal segments while the EEG signals on other lead channels do not have the waveform characteristics related to epileptic seizure discharges on this candidate signal segment. Therefore, the failure judgment can not only screen out the failed lead channels, but also further screen out the failed candidate signal sub - segments.

[0055] After screening out the failed lead channels and the invalid candidate signal sub - segments, the energy correlation value and the waveform correlation value are calculated in sub - steps S143 and S144 respectively. Sub - steps S143 and S144 can be executed in parallel or in any order, Figure 1 and an exemplary execution manner of the two is shown in a parallel manner.

[0056] In sub-step S143, for any two lead channels among the multiple lead channels of the EEG signals determined to be valid, calculate the inter-channel energy correlation value between the candidate signal segments (if sub-steps S141 and S142 do not exist) or candidate signal sub-segments (if sub-steps S141 and S142 exist) on these two lead channels, and determine the energy correlation value of the candidate signal segment or candidate signal sub-segment based on the inter-channel energy correlation values calculated for all combinations of two lead channels among the multiple lead channels.

[0057] The inter-channel energy correlation value of the EEG signal can be calculated, for example, by the following formula (1):

[0058]

[0059] Wherein, v1 and v2 respectively represent the signal vectors of the overlapping parts when the EEG signals of the candidate signal segments or candidate signal sub-segments on the two lead channels slide on the time axis. length represents the duration length of the overlapping parts of the EEG signals on the two lead channels. It can be understood that the vectors v1 and v2 are respectively the signal vectors of the EEG signals on the two lead channels, and their vector lengths are actually the number of sampling points in the candidate signal segments or candidate signal sub-segments, and the vector directions are consistent with the direction of the time axis. The vector multiplication v1*v2 can be calculated as the sum of the products of the EEG signal waveform amplitudes of each sampling point in the overlapping parts when the EEG signals on the two lead channels slide simultaneously on the time axis. The inter-channel energy correlation value corr is the maximum value of the quotient obtained by dividing the product of the signal vectors v1*v2 in the overlapping part between the two lead channels by the duration length (number of sampling points) of the overlapping part.

[0060] After calculating the inter-channel energy correlation values corr corresponding to all combinations of pairwise lead channels of the candidate signal segments or candidate signal sub-segments, calculate the total energy correlation value based on these inter-channel energy correlation values. According to the embodiments of the present application, the energy correlation value of the candidate signal segment or candidate signal sub-segment can be calculated based on the average value of these inter-channel energy correlation values. It is also possible to sort these inter-channel energy correlation values according to their magnitudes, and select the average value of multiple higher numerical inter-channel energy correlation values to calculate the total energy correlation value. For example, the average value can be calculated by selecting the inter-channel energy correlation values ranked in the top ten percent or the top 5 in terms of the inter-channel energy correlation values. According to the embodiments of the present application, when a candidate signal segment has a long duration and is segmented into several candidate signal sub-segments, the energy correlation value of the candidate signal segment can be calculated by the energy correlation values of each sub-candidate signal segment. For example, the maximum value of the energy correlation values of the sub-candidate signal segments is used as the energy correlation value of the candidate signal segment.

[0061] In sub-step S144, for any two lead channels among the multiple lead channels of the EEG signals determined to be valid, calculate the inter-channel waveform correlation value between the candidate signal segments (if sub-steps S141 and S142 do not exist) or candidate signal sub-segments (if sub-steps S141 and S142 exist) on these two lead channels, and determine the waveform correlation value of the candidate signal segment or candidate signal sub-segment based on the inter-channel waveform correlation values calculated for all combinations of two lead channels among the multiple lead channels.

[0062] The inter-channel waveform correlation value of the EEG signal can be calculated, for example, by the following formula (2):

[0063]

[0064] Similar to formula (1), here v1 and v2 respectively represent the signal vectors of the overlapping parts when the EEG signals of the candidate signal segments or candidate signal sub-segments on the two lead channels slide on the time axis. norm(v) represents taking the norm of the signal vector v. The vector multiplication v1*v2 can also be calculated as the sum of the products of the EEG signal waveform amplitudes at each sampling point of the overlapping part when the EEG signals on the two lead channels slide simultaneously on the time axis. The norm operation in the denominator can be performed by taking the L2 norm of the signal vector, so that the norm units of the two signal vectors calculated in the denominator are consistent, thereby eliminating the influence of the signal amplitude value on the denominator calculation. In this way, the product of the norms in the denominator can be calculated by scalar multiplication. It can be understood that the quotient of the product of the signal vectors in the numerator divided by the product of the norms of the signal vectors in the denominator is equivalent to normalizing the vector product, so that the calculation of the inter-channel waveform correlation value does not consider the magnitude of the EEG signal waveform but only the similarity of the waveforms. The value of the quotient inside the brackets finally calculated should be within the range of [0,1]. The inter-channel waveform correlation value sim is the maximum value of the above quotient of the signal vector v1*v2 in the overlapping part between the two lead channels.

[0065] After calculating the inter-channel waveform correlation values sim corresponding to all pairwise combinations of lead channels of the candidate signal segment or candidate signal sub-segments, the total waveform correlation value is calculated based on these inter-channel waveform correlation values. Similar to sub-step S143, according to an embodiment of the present application, the waveform correlation value of the candidate signal segment or candidate signal sub-segment can be calculated based on the average value of these inter-channel waveform correlation values. It is also possible to sort these inter-channel waveform correlation values by magnitude, and select the average value of multiple inter-channel waveform correlation values with higher numerical values to calculate the total waveform correlation value. For example, the average value can be calculated by selecting the top ten percent or the top 5 inter-channel waveform correlation values. According to an embodiment of the present application, when a candidate signal segment has a long duration and is segmented into several candidate signal sub-segments, the waveform correlation value of the candidate signal segment can be calculated through the waveform correlation values of each sub-candidate signal segment. For example, the maximum value of the waveform correlation values of the sub-candidate signal segments is used as the waveform correlation value of the candidate signal segment.

[0066] After separately calculating the energy correlation value and the waveform correlation value of the candidate signal segment or candidate signal sub-segment, in sub-step S145, the method separately compares the energy correlation value and the waveform correlation value with their respective preset thresholds to determine whether there is a real epileptic seizure in the candidate signal segment or candidate signal sub-segment, or whether it indicates a real epileptic seizure. The energy correlation threshold and the waveform correlation threshold are set based on the statistical results of the energy correlation and waveform correlation data in the EEG signals of different epileptic patients during the epileptic seizure period. The comparison of the energy correlation value with the energy correlation threshold and the comparison of the waveform correlation value with the waveform correlation threshold are equally important. A real epileptic seizure can be determined only when both of these two judgment conditions are met, because the judgment criteria should not only ensure the synchrony of the waveform amplitude but also exclude the adverse effects brought by the high synchrony caused by too low waveform amplitude.

[0067] As described above, if the analysis in step S140 is performed on at least one of the extracted high-frequency EEG signal and low-frequency EEG signal, then at least one of the high-frequency and low-frequency EEG signals needs to meet the judgment conditions of the energy correlation value and the waveform correlation value, and different energy correlation thresholds and waveform correlation thresholds can be set for the low-frequency and high-frequency EEG signals.

[0068] For Figure 5A the candidate signal segment 501 to be analyzed in the EEG waveform diagram shown in, since at least one of its energy correlation value and waveform correlation value does not meet the above judgment conditions, it is identified as not a real epileptic seizure, that is, it belongs to the signal segment during the inter-ictal period. And Figure 5BThe energy correlation value and waveform correlation value of the candidate signal segment 502 to be analyzed therein both meet the above judgment conditions, and thus it is identified as a real epileptic seizure.

[0069] Finally, the signal segments of the long-duration type and the short-duration type that are determined to have epileptic seizures are combined (i.e., their union) to obtain all the EEG signal segments that are determined to have epileptic seizures.

[0070] Figure 6 Fig. 6 shows a device 600 for detecting epileptic seizures according to an embodiment of the present application, which includes an acquisition unit 610 and a detection unit 620. The acquisition unit 610 is used to perform Figure 1 the steps S110 shown in Fig. 7 to acquire the electroencephalogram signals of the subject on multiple lead channels. The detection unit 620 is used to perform Figure 1 the steps S120 to S140 shown in Fig. 8, at least including extracting candidate signal segments from multiple signal segments based on the time characteristics and frequency characteristics of the EEG signals, and analyzing the correlation indexes of the candidate signal segments to determine whether the candidate signal segments indicate epileptic seizures. Optionally, the EEG signals can also be preprocessed before extracting the candidate signal segments. The detection unit 620 can also perform Figure 1 more detailed operations of the sub-steps of the steps described in Fig. 9, which will not be elaborated here.

[0071] By adopting the epileptic seizure detection solution proposed in the embodiment of the present application, suspicious candidate signal segments are extracted respectively based on the frequency characteristics and time characteristics of the EEG signals, where the characteristics of prominent low-frequency and high-frequency signal energies in different epileptic seizure types are emphasized, and the duration lengths of different types of epileptic seizure conditions are specifically analyzed. Further, the correlation indexes of the EEG signals on multiple lead channels are analyzed during the extraction of the candidate signal segments to screen out the accurate signal segments indicating epileptic seizures. This solution can identify epileptic seizures from the EEG signals of multiple information channels with a more refined detection standard. In particular, it can effectively distinguish the signal characteristics between the EEG signals during epileptic seizure discharges and the EEG signals with artifacts. It can be applied to any EEG signal containing abnormal discharge waveforms during epileptic seizures, and the signal characteristics of the signal segments are examined based on the duration of different types of epileptic seizure discharges without the need for continuous long-term monitoring of the EEG signals, thereby obtaining a detection result with higher accuracy and sensitivity.

[0072] It should be noted that although several modules or units of a system for detecting epileptic seizures are mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present application, the features and functions of two or more of the above-described modules or units can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units. The components shown as modules or units may or may not be physical units, that is, they may be located in one place or distributed over multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of the present application. A person of ordinary skill in the art can understand and implement it without creative work.

[0073] In an exemplary embodiment of the present application, there is also provided a computer-readable storage medium having a computer program stored thereon, the program including executable instructions that, when executed by, for example, a processor, can implement the steps of the method for detecting epileptic seizures described in any of the above embodiments. In some possible implementation manners, various aspects of the present application can also be implemented in the form of a program product, which includes program code that, when the program product runs on a terminal device, causes the terminal device to execute the steps described in the method for detecting epileptic seizures in this specification according to various exemplary embodiments of the present application.

[0074] The program product for implementing the above method according to the embodiments of the present application can be a portable compact disc read-only memory (CD-ROM) and includes program code and can run on a terminal device, such as a personal computer. However, the program product of the present application is not limited to this. In this document, a readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0075] The program product can adopt any combination of one or more readable media. The readable media can be a readable signal medium or a readable storage medium. A readable storage medium can, for example, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0076] The computer-readable storage medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries the readable program code. Such a propagated data signal may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. The readable storage medium may also be any readable medium other than the readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0077] The program code for performing the operations of the present application may be written in any combination of one or more programming languages. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, executed as a stand-alone software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computing device (e.g., by connecting through the Internet using an Internet service provider).

[0078] In an exemplary embodiment of the present application, an electronic device is further provided. The electronic device may include a processor and a memory for storing executable instructions of the processor. Among them, the processor is configured to execute the steps of the method for detecting epileptic seizures in any one of the foregoing embodiments by executing the executable instructions.

[0079] Those skilled in the art can understand that various aspects of the present application can be implemented as a system, method, or program product. Therefore, various aspects of the present application can be specifically implemented in the following forms, namely: a complete hardware implementation, a complete software implementation (including firmware, microcode, etc.), or an implementation combining hardware and software aspects, which can be collectively referred to as "circuitry", "module", or "system" here.

[0080] The following refers to Figure 7 to describe the electronic device 700 according to this embodiment of the present application. Figure 7 The illustrated electronic device 700 is merely an example and should not impose any limitation on the functions and usage scope of the embodiments of the present application.

[0081] As Figure 7As shown, the electronic device 700 is presented in the form of a general-purpose computing device. The components of the electronic device 700 may include, but are not limited to: at least one processing unit 710, at least one storage unit 720, a bus 730 connecting different system components (including the storage unit 720 and the processing unit 710), a display unit 740, etc.

[0082] Among them, the storage unit stores program code, and the program code can be executed by the processing unit 710, so that the processing unit 710 executes the steps according to various exemplary embodiments of the present application described in the method for detecting epileptic seizures in this specification. For example, the processing unit 710 can execute as Figure 1 the steps shown in.

[0083] The storage unit 720 may include a readable medium in the form of a volatile storage unit, such as a random access storage unit (RAM) 7201 and / or a cache storage unit 7202, and may further include a read-only storage unit (ROM) 7203.

[0084] The storage unit 720 may further include a program / utility 7204 having a set (at least one) of program modules 7205. Such program modules 7205 include, but are not limited to: an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include the implementation of a network environment.

[0085] The bus 730 may represent one or more of several types of bus structures, including a storage unit bus or a storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any bus structure in a variety of bus structures.

[0086] The electronic device 700 can also communicate with one or more external devices 800 (such as a keyboard, a pointing device, a Bluetooth device, etc.), and can also communicate with one or more devices that enable a user to interact with the electronic device 700, and / or communicate with any device that enables the electronic device 700 to communicate with one or more other computing devices (such as a router, a modem, etc.). Such communication can be carried out through the input / output (I / O) interface 750. Moreover, the electronic device 700 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through the network adapter 760. The network adapter 760 can communicate with other modules of the electronic device 700 through the bus 730. It should be understood that although not shown in the figure, other hardware and / or software modules can be used in combination with the electronic device 700, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0087] Through the description of the above embodiments, those skilled in the art can easily understand that the exemplary embodiments described herein can be implemented by software, or can be implemented by the way of software combined with necessary hardware. Therefore, the technical solutions according to the embodiments of the present application can be embodied in the form of a software product, and the software product can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.) or on a network, including several instructions to enable a computing device (which can be a personal computer, a server, or a network device, etc.) to execute the method for detecting epileptic seizures according to the embodiments of the present application.

[0088] After considering the specification and practicing the content disclosed herein, those skilled in the art will easily think of other implementation schemes of the present application. The present application aims to cover any variations, uses, or adaptive changes of the present application, and these variations, uses, or adaptive changes follow the general principles of the present application and include the common general knowledge or conventional technical means in the technical field not disclosed in the present application. The specification and the embodiments are only regarded as exemplary, and the true scope and spirit of the present application are pointed out by the appended claims.

Claims

1. A method for detecting epileptic seizures, comprising: Obtaining electroencephalogram (EEG) signals of a subject on multiple lead channels, wherein the EEG signals have multiple signal segments; Extracting candidate signal segments from the multiple signal segments based on the temporal features and frequency features of the EEG signals, wherein In the multiple signal segments, extracting long-duration type signal segments and short-duration type signal segments corresponding to epileptic seizures of different duration lengths; For the long-duration type and short-duration type signal segments respectively, extracting the candidate signal segments from high-frequency EEG signals and low-frequency EEG signals; Analyzing a correlation index of the candidate signal segments to determine whether the candidate signal segments indicate epileptic seizures, wherein the correlation index includes an energy correlation value and a waveform correlation value, and wherein For the long-duration type and short-duration type signal segments respectively, calculating the energy correlation value and the waveform correlation value of the candidate signal segments in the high-frequency EEG signals and the low-frequency EEG signals; Determining the candidate signal segments with the energy correlation value exceeding an energy correlation threshold and the waveform correlation value exceeding a waveform correlation threshold in at least one of the high-frequency EEG signals and the low-frequency EEG signals as indicating epileptic seizures.

2. The method according to claim 1, wherein Extracting candidate signal segments from high-frequency EEG signals and low-frequency EEG signals further includes: Extracting the high-frequency EEG signals and the low-frequency EEG signals of the EEG signals; For the high-frequency EEG signals and the low-frequency EEG signals respectively, screening signal segments whose waveform amplitudes exceed waveform amplitude thresholds corresponding to different signal duration lengths; Determining the candidate signal segments based on the screened signal segments.

3. The method according to claim 2, characterized in that, Determining the candidate signal segments based on the screened signal segments includes: Merging the screened signal segments adjacent in time, and extracting the merged signal segments whose signal durations meet the time range conditions corresponding to long-duration epileptic seizures and short-duration epileptic seizures respectively as the candidate signal segments.

4. The method according to claim 1, wherein Calculating the energy correlation value of the candidate signal segments includes: Calculating an inter-channel energy correlation value between the candidate signal segments on any two of the multiple lead channels; Determining the energy correlation value of the candidate signal segments based on all the inter-channel energy correlation values.

5. The method according to claim 4, characterized in that, Calculating the inter-channel energy correlation value based on the maximum value of the quotient of the product of the signal vectors of the overlapping parts when the candidate signal segments on two lead channels slide on the time axis divided by the length of the duration of the overlapping parts.

6. The method according to claim 4, characterized in that, Determining the energy correlation value of the candidate signal segments based on the average value of all the inter-channel energy correlation values.

7. The method according to claim 6, characterized in that Determining the energy correlation value of the candidate signal segments based on the average value of multiple inter-channel energy correlation values with higher numerical values among all the inter-channel energy correlation values.

8. The method according to claim 1, wherein Calculating the waveform correlation value of the candidate signal segments includes: Calculating an inter-channel waveform correlation value between the candidate signal segments on any two of the multiple lead channels; Determining the waveform correlation value of the candidate signal segments based on all the inter-channel waveform correlation values.

9. The method according to claim 8, wherein Calculate the inter-channel waveform correlation value based on the maximum value of the quotient obtained by dividing the product of the signal vectors of the overlapping parts when the candidate signal segments on two lead channels slide on the time axis by the product of the magnitudes of the signal vectors.

10. The method according to claim 9, characterized in that, Determine the waveform correlation value of the candidate signal segment based on the average value of all the inter-channel waveform correlation values.

11. The method according to claim 1, wherein Analyzing the correlation index of the candidate signal segment to determine whether the candidate signal segment indicates a seizure further includes: Dividing the candidate signal segment into candidate signal sub-segments with a duration of 3 seconds; Analyzing the correlation index for the candidate signal sub-segment to determine whether the candidate signal sub-segment indicates a seizure.

12. The method according to claim 11, wherein Analyzing the correlation index of the candidate signal segment to determine whether the candidate signal segment indicates a seizure further includes: Determining the correlation index of the candidate signal segment based on the correlation indices of the candidate signal sub-segments included in the candidate signal segment to determine whether the candidate signal segment indicates a seizure.

13. The method according to claim 12, wherein Determining the correlation index of the candidate signal segment based on the correlation indices of the candidate signal sub-segments included in the candidate signal segment includes: Taking the maximum value among the energy correlation values of the candidate signal sub-segments as the energy correlation value of the candidate signal segment; and / or Taking the maximum value among the waveform correlation values of the candidate signal sub-segments as the waveform correlation value of the candidate signal segment.

14. The method according to claim 11, wherein Analyzing the correlation index of the candidate signal segment to determine whether the candidate signal segment indicates a seizure further includes: Before analyzing the correlation index of the candidate signal sub-segment, screening out invalid lead channels and / or invalid candidate signal sub-segments based on the comparison between the absolute value of the waveform amplitude of the candidate signal sub-segment and the artifact waveform amplitude threshold.

15. The method according to claim 1, characterized in that, Further includes: Before extracting the candidate signal segment, performing preprocessing operations on the electroencephalogram signal including at least one of the following: Resampling the electroencephalogram signal; Removing the electroencephalogram signal on invalid lead channels; Removing power frequency noise, baseline drift noise, and / or high-frequency noise in the electroencephalogram signal; Calibrating the electroencephalogram signal based on a reference signal.

16. A device for detecting seizures, comprising: An acquisition unit configured to acquire electroencephalogram signals of a subject on multiple lead channels, the electroencephalogram signals having multiple signal segments; A detection unit configured to extract candidate signal segments from the multiple signal segments based on the time characteristics and frequency characteristics of the electroencephalogram signals, and analyze the correlation indices of the candidate signal segments to determine whether the candidate signal segments indicate seizures, wherein extracting candidate signal segments from the multiple signal segments based on the time characteristics and frequency characteristics of the electroencephalogram signals includes: Extracting long-duration type signal segments and short-duration type signal segments corresponding to seizures of different duration lengths from the multiple signal segments; For the long-duration type and short-duration type signal segments respectively, extracting the candidate signal segments from the high-frequency electroencephalogram signals and low-frequency electroencephalogram signals; Among them, the correlation index includes an energy correlation value and a waveform correlation value. Analyzing the correlation index of the candidate signal segment to determine whether the candidate signal segment indicates a seizure includes: Calculating the energy correlation value and the waveform correlation value of the candidate signal segment in the high-frequency EEG signal and the low-frequency EEG signal respectively for the signal segments of the long-duration type and the short-duration type; Determining the candidate signal segments in which the energy correlation value in at least one of the high-frequency EEG signal and the low-frequency EEG signal exceeds the energy correlation threshold and the waveform correlation value exceeds the waveform correlation threshold as indicating a seizure.

17. A computer-readable storage medium, on which a computer program is stored, the computer program including executable instructions, which when executed by a processor, implement the method according to any one of claims 1 to 15.

18. An electronic device, characterized in that, Including: A processor; And A memory for storing the executable instructions of the processor; Among them, the executable instructions, when executed by the processor, implement the method according to any one of claims 1 to 15.

Images