Epileptic discharge detection method, device, program product and electronic device

Abstract

The present disclosure provides an epileptic discharge detection method, device, program product and electronic device, which relate to the field of computer technology. The method includes: obtaining biomedical detection data of the subject; extracting a first electrical signal feature and a first sleep feature from the biomedical detection data; and determining whether the subject has an epileptic discharge according to the first electrical signal feature and the first sleep feature. The present disclosure improves the accuracy and reliability of epileptic discharge detection results and provides doctors with effective auxiliary diagnosis and treatment information. It also reduces the workload of manual post-processing and improves detection efficiency.

Description

Technical Field

[0001] The embodiments of the present disclosure relate to the field of computer technology, and more specifically, the embodiments of the present disclosure relate to an epileptic discharge detection method, an epileptic discharge detection device, a computer program product and an electronic device. Background Art

[0002] This section is intended to provide a background or context to the embodiments of the disclosure that are recited in the claims, and no description herein is admitted to be prior art by inclusion in this section.

[0003] Epilepsy is a common neurological disease. Detecting epileptic discharges in medical test data such as electroencephalograms is one of the important criteria for diagnosing epilepsy.

[0004] Faced with the increasing demand for relevant data analysis and detection, the manual method of analyzing data and identifying epileptic discharges is inefficient, and the manual identification results usually have low consistency and are difficult to guarantee reliability. Therefore, the industry has an increasing demand for automated detection of epileptic discharges.

[0005] In the related art, artificial intelligence technology is used to detect epileptic discharges. For example, by training a neural network for epileptic discharge recognition, it is possible to automatically process electroencephalogram and other detection data and output epileptic discharge detection results. Summary of the invention

[0006] However, the current AI-based epilepsy discharge detection method has low accuracy and often requires professional doctors and technicians to post-process the AI ​​detection results in order to obtain accurate evaluation results, which affects detection efficiency.

[0007] In view of the above problems, embodiments of the present disclosure provide an epileptic discharge detection method, an epileptic discharge detection device, a computer program product, and an electronic device.

[0008] According to a first aspect of the present disclosure, a method for detecting epileptic discharges is provided, the method comprising: acquiring biomedical detection data of a test subject; extracting a first electrical signal feature and a first sleep feature from the biomedical detection data; and determining whether epileptic discharges occur in the test subject based on the first electrical signal feature and the first sleep feature.

[0009] In one embodiment, extracting the first electrical signal feature and the first sleep feature from the biomedical detection data includes: inputting the biomedical detection data into a first feature network to obtain a first feature tensor; the first feature tensor includes the first electrical signal feature and the first sleep feature; wherein the first feature network and the second feature network share at least some parameters; the first feature network is a network used to extract features in an epileptic discharge detection task, and the second feature network is a network used to extract features in a sleep state detection task.

[0010] In one embodiment, determining whether the subject has epileptic discharges based on the first electrical signal feature and the first sleep feature includes: inputting the first feature tensor into a first classification network to obtain a first classification result; the first classification result indicates whether the subject has epileptic discharges.

[0011] In one embodiment, the method also includes: inputting the sample biomedical detection data into the first feature network to obtain a first sample feature tensor; the first sample feature tensor includes a first sample electrical signal feature and a first sample sleep feature; inputting the first sample feature tensor into the first classification network to obtain a first sample classification result; inputting the biomedical detection data into the second feature network to obtain a second sample feature tensor; the second sample feature tensor includes a second sample electrical signal feature and a second sample sleep feature; inputting the second sample feature tensor into the second classification network to obtain a second sample classification result; updating the first feature network and the first classification network according to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label; the first label is the epileptic discharge label corresponding to the sample biomedical detection data, and the second label is the sleep state label corresponding to the sample biomedical detection data.

[0012] In one embodiment, the method further includes: updating the second feature network and the second classification network according to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label.

[0013] In one embodiment, the method also includes: inputting the biomedical detection data into the second feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; inputting the second feature tensor into a second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0014] In one embodiment, extracting the first electrical signal feature and the first sleep feature from the biomedical detection data includes: inputting the biomedical detection data into a third feature network to obtain basic features; inputting the basic features into a fourth feature network to obtain a first feature tensor; the first feature tensor includes the first electrical signal feature and the first sleep feature; wherein the third feature network is a network for extracting basic features in epileptic discharge detection tasks and sleep state detection tasks.

[0015] In one embodiment, the method also includes: inputting the basic features into a fifth feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; inputting the second feature tensor into a second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0016] In one embodiment, the obtaining of biomedical test data of the test subject includes: preprocessing the original test data of the test subject to obtain the biomedical test data; the preprocessing includes one or more of the following processes: resampling, filtering, data enhancement, noise data removal, numerical normalization, and time-frequency transformation.

[0017] According to a second aspect of the present disclosure, an epileptic-like discharge detection device is provided, the device comprising: a data acquisition module, configured to acquire biomedical detection data of a test subject; a feature extraction module, configured to extract a first electrical signal feature and a first sleep feature from the biomedical detection data; and a classification module, configured to determine whether the test subject has an epileptic-like discharge based on the first electrical signal feature and the first sleep feature.

[0018] In one embodiment, extracting the first electrical signal feature and the first sleep feature from the biomedical detection data includes: inputting the biomedical detection data into a first feature network to obtain a first feature tensor; the first feature tensor includes the first electrical signal feature and the first sleep feature; wherein the first feature network and the second feature network share at least some parameters; the first feature network is a network used to extract features in an epileptic discharge detection task, and the second feature network is a network used to extract features in a sleep state detection task.

[0019] In one embodiment, determining whether the subject has epileptic discharges based on the first electrical signal feature and the first sleep feature includes: inputting the first feature tensor into a first classification network to obtain a first classification result; the first classification result indicates whether the subject has epileptic discharges.

[0020] In one embodiment, the device also includes a training module, which is configured to: input the sample biomedical detection data into the first feature network to obtain a first sample feature tensor; the first sample feature tensor includes a first sample electrical signal feature and a first sample sleep feature; input the first sample feature tensor into the first classification network to obtain a first sample classification result; input the biomedical detection data into the second feature network to obtain a second sample feature tensor; the second sample feature tensor includes a second sample electrical signal feature and a second sample sleep feature; input the second sample feature tensor into the second classification network to obtain a second sample classification result; update the first feature network and the first classification network according to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label; the first label is the epileptic discharge label corresponding to the sample biomedical detection data, and the second label is the sleep state label corresponding to the sample biomedical detection data.

[0021] In one embodiment, the training module is further configured to update the second feature network and the second classification network according to the difference between the first sample classification result and the first label and the difference between the second sample classification result and the second label.

[0022] In one embodiment, the feature extraction module is further configured to: input the biomedical detection data into the second feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; the classification module is further configured to: input the second feature tensor into a second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0023] In one embodiment, extracting the first electrical signal feature and the first sleep feature from the biomedical detection data includes: inputting the biomedical detection data into a third feature network to obtain basic features; inputting the basic features into a fourth feature network to obtain a first feature tensor; the first feature tensor includes the first electrical signal feature and the first sleep feature; wherein the third feature network is a network for extracting basic features in epileptic discharge detection tasks and sleep state detection tasks.

[0024] In one embodiment, the feature extraction module is further configured to: input the basic feature into a fifth feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; the classification module is further configured to: input the second feature tensor into a second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0025] In one embodiment, the obtaining of biomedical test data of the test subject includes: preprocessing the original test data of the test subject to obtain the biomedical test data; the preprocessing includes one or more of the following processes: resampling, filtering, data enhancement, noise data removal, numerical normalization, and time-frequency transformation.

[0026] According to a third aspect of the present disclosure, a computer program product is provided, including a computer program, wherein when the computer program is executed by a processor, the method of the first aspect and possible implementations thereof are implemented.

[0027] According to a fourth aspect of the present disclosure, an electronic device is provided, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the method of the above-mentioned first aspect and its possible implementation methods by executing the executable instructions.

[0028] The embodiments of the present disclosure have the following technical effects:

[0029] In the task of detecting epileptic discharges, in addition to extracting the first electrical signal feature, the first sleep feature is also extracted, and whether the subject has epileptic discharges is determined based on the first electrical signal feature and the first sleep feature. The first sleep feature provides the sleep information of the subject, and by combining with the first electrical signal feature, it can help the computer distinguish similar epileptic discharges from sleep physiological waves, thereby reducing the interference of sleep physiological waves on epileptic discharge detection, improving the accuracy and reliability of epileptic discharge detection results, and providing doctors with effective auxiliary diagnosis and treatment information. It also reduces the workload of manual post-processing and improves detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1A A schematic diagram showing a system architecture in an embodiment of the present disclosure.

[0031] Figure 1B A schematic diagram showing another system architecture in an embodiment of the present disclosure.

[0032] Figure 2 A flow chart of an epileptic discharge detection method in an embodiment of the present disclosure is shown.

[0033] Figure 3 A schematic diagram of an electroencephalogram in an embodiment of the present disclosure is shown.

[0034] Figure 4 A flow chart for obtaining a second classification result in an embodiment of the present disclosure is shown.

[0035] Figure 5 A flowchart of a training model in an embodiment of the present disclosure is shown.

[0036] Figure 6 A schematic diagram showing annotated data in an embodiment of the present disclosure.

[0037] Figure 7 A schematic diagram showing a solution architecture in an embodiment of the present disclosure.

[0038] Figure 8 A schematic structural diagram of an epileptic discharge detection device in an embodiment of the present disclosure is shown.

[0039] Fig. 9 A schematic structural diagram of an electronic device in an embodiment of the present disclosure is shown.

[0040] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0041] The following is a detailed description of the principles and spirit of the present disclosure with reference to several representative embodiments of the present disclosure. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present disclosure, and are not intended to limit the scope of the present disclosure in any way. On the contrary, these embodiments are provided to make the present disclosure more thorough and complete, and to fully convey the scope of the present disclosure to those skilled in the art.

[0042] The embodiments of the present disclosure may be implemented as a system, device, apparatus, method or computer program product. Therefore, the present disclosure may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software. SUMMARY OF THE INVENTION

[0044] In the related art, artificial intelligence technology is used to detect epileptic discharges. For example, by training a neural network for epileptic discharge recognition, it is possible to automatically process electroencephalogram and other detection data and output epileptic discharge detection results.

[0045] However, the current AI-based epileptiform discharge detection method has low accuracy and often requires professional doctors and technicians to post-process the AI ​​detection results, affecting detection efficiency.

[0046] The inventors have found that sleep physiological waves in EEG and other detection data, such as top spikes and K complex waves, are similar to the characteristics of epileptic discharges, and can easily interfere with the detection of epileptic discharges, resulting in inaccurate detection results.

[0047] In view of the above, the present disclosure provides an epileptic discharge detection method, an epileptic discharge detection device, a computer program product and an electronic device. In the epileptic discharge detection task, in addition to extracting the first electrical signal feature, the first sleep feature is also extracted, and whether the subject has an epileptic discharge is determined based on the first electrical signal feature and the first sleep feature. The first sleep feature provides the sleep information of the subject, and by combining with the first electrical signal feature, it can help the computer to distinguish similar epileptic discharges from sleep physiological waves, thereby reducing the interference of sleep physiological waves on epileptic discharge detection, improving the accuracy and reliability of epileptic discharge detection results, and providing doctors with effective auxiliary diagnosis and treatment information. It also reduces the workload of manual post-processing and improves detection efficiency.

[0048] After introducing the basic principles of the present disclosure, various non-limiting embodiments of the present disclosure are described in detail below.

[0049] Application Scenario Overview

[0050] It should be noted that the following application scenarios are only shown to facilitate understanding of the spirit and principles of the present disclosure, and the implementation methods of the present disclosure are not limited in this regard and can be applied to any applicable scenarios.

[0051] The embodiments of the present disclosure can be applied to the detection of suspected patients and related scenarios of auxiliary diagnosis and treatment. For example, if a patient is suspected of having epilepsy, the patient's electroencephalogram and other biomedical test data can be tested, and the epileptiform discharge detection method disclosed in the present disclosure can be used to determine whether the patient has epileptiform discharges. The test results can assist doctors in diagnosing patients. The following is a specific description of the application scenario in conjunction with the system architecture.

[0052] Figure 1A A schematic diagram of a system architecture for epileptiform discharge detection is shown. The system architecture includes a test object 101, a biomedical monitoring device 102, and a processing device 103. When it is necessary to perform epileptiform discharge detection on the test object 101, the biomedical monitoring device 102 can be set for the test object 101 to collect biomedical detection data, and the data is processed by the processing device 103.

[0053] The biomedical monitoring device 102 may include detection electrodes 1021, 1022, 1023 and a main device 1024. The detection electrodes 1021, 1022, 1023 may contact different parts of the subject 101 to collect electrical signals, and the main device 1024 aggregates and processes the electrical signals to obtain original detection data or pre-processed biomedical detection data. Exemplarily, the biomedical monitoring device 102 may be an electroencephalogram (EEG) monitoring device (such as an electroencephalograph), and the detection electrodes 1021, 1022, 1023 may be fixed at different positions on the head of the subject 101 to collect multi-channel EEG signals. Of course, Figure 1A The biomedical monitoring device 102 shown is merely exemplary and may also include a helmet, a hat, bedding, packaged portable detection electrodes, and other components, which are not limited in the present disclosure.

[0054] The biomedical monitoring device 102 can be connected to the processing device 103 in communication, such as through a wired or wireless communication link, so as to send the collected data to the processing device 103. The processing device 103 can execute the epileptic discharge detection method in this exemplary embodiment, process the acquired biomedical detection data, and obtain the epileptic discharge detection result of the subject 101. In one embodiment, the processing device 103 may include a display, which can display the epileptic discharge detection result, and can also display the original detection data or biomedical detection data.

[0055] In one embodiment, the biomedical monitoring device 102 and the processing device 103 may be integrated into the same device. Exemplarily, the main device 1024 of the biomedical monitoring device 102 may serve as the processing device 103 .

[0056] Figure 1B A schematic diagram of another system architecture for epileptiform discharge detection is shown. The system architecture includes a subject 101, a biomedical monitoring device 102, a data transceiver device 104, and a server 105. The biomedical monitoring device 102 can be connected to the data transceiver device 104 for communication, and the data transceiver device 104 is connected to the server 105 for communication. Thus, after the data transceiver device 104 obtains the original detection data or the biomedical detection data, it sends it to the server 105. The server 105 may include any form of data processing server such as a cloud server and a distributed server. After obtaining the data sent by the data transceiver device 104, the server 105 obtains the epileptiform discharge detection result of the subject 101 by executing the epileptiform discharge detection method in the embodiment of the present disclosure. In one embodiment, the server 105 can return the epileptiform discharge detection result to the data transceiver device 104 for display on the data transceiver device 104 or the biomedical monitoring device 102.

[0057] Figure 1B The system architecture shown is suitable for portable scenarios. For example, the subject 101 can use the portable biomedical monitoring device 102 to collect data at home, office, etc., and the data transceiver device 104 sends the data to the server 105 for epileptiform discharge detection. In this way, the user can learn the epileptiform discharge detection results without leaving home. When encountering suspected epilepsy and other problems, the obtained epileptiform discharge detection results can be sent to the doctor to help the doctor judge the condition.

[0058] Exemplary Methods

[0059] The following is an exemplary description of the epileptic discharge detection method in the embodiment of the present disclosure. Figure 2 The process of the epileptiform discharge detection method is shown, which may include steps S210 to S230. Figure 2 Provide detailed instructions for each step.

[0060] refer to Figure 2 In step S210, biomedical test data of the subject is obtained.

[0061] Among them, the subject refers to a person who needs to undergo epileptiform discharge detection, such as a possible epilepsy patient. Biomedical detection data is data detected from the human body through medical detection methods, which can reflect the physiological state or physical signs of the person. Biomedical detection data includes but is not limited to any one or more of the following data: EEG data, ECG data, EMG data, EO data, GEG data, etc. It can also include non-electrical data such as body temperature, blood pressure, pulse, and respiration. The following is an exemplary description of biomedical detection data and its collection process.

[0062] EEG data can be an EEG collected by a multi-channel EEG machine. An EEG (Electroencephalogram) is a graph obtained by amplifying and recording the spontaneous biopotential of the cerebral cortex from the scalp through a sophisticated instrument. It is the spontaneous and rhythmic electrical activity of a group of brain cells recorded by electrodes. This electrical activity is a plane diagram of the relationship between the potential and time, with the potential as the vertical axis and time as the horizontal axis. The frequency (period), amplitude and phase of brain waves constitute the basic characteristics of an EEG.

[0063] ECG data can be an electrocardiogram collected by a multi-lead electrocardiogram machine. Electrocardiogram (ECG) refers to the graph of various forms of potential changes drawn from the body surface by an electrocardiograph during each cardiac cycle, with the pacemaker, atrium, and ventricle being excited successively, accompanied by changes in the electrocardiogram bioelectricity. The electrocardiogram is an objective indicator of the occurrence, propagation, and recovery process of cardiac excitement.

[0064] Myoelectric data (such as electromyography (EMG)) is the temporal and spatial superposition of action potentials of motor units in many muscle fibers, which can be obtained by sticking electromyographic sensors on the skin.

[0065] Electrooculogram (EOG) data, such as electrooculogram (EOG), is a bioelectric signal caused by the potential difference between the cornea and retina of the eye. It is very easy to collect and can be completed with a small number of electrodes.

[0066] Gastrointestinal data (such as electrogastrogram (EGG)) are electrical signals generated by the contraction of stomach muscles and can be collected on the surface of the abdominal skin using electrodes.

[0067] In one embodiment, the above-mentioned acquisition of biomedical test data of the subject may include the following steps:

[0068] Preprocessing the original test data of the tested object to obtain biomedical test data;

[0069] Preprocessing includes one or more of the following: resampling, filtering, data enhancement, noise data removal, numerical standardization, and time-frequency transformation.

[0070] The original detection data is the original data collected by the biomedical monitoring equipment. For example, the biomedical monitoring equipment includes an electroencephalograph, and the original detection data includes the original EEG data or electroencephalogram collected by the electroencephalogram. The electroencephalogram is essentially a graph drawn by the EEG signal data at different times, and the EEG signal can be considered to be equivalent to the electroencephalogram. Figure 3 The schematic diagram of multi-channel EEG data is shown. The original EEG data is plotted into a graph to obtain Figure 3 The waveform shown is the electroencephalogram.

[0071] To facilitate subsequent processing, the original detection data may be preprocessed. The specific method of preprocessing is exemplified below.

[0072] ① Resampling. Resampling can change the sampling rate of the signal and can convert non-uniformly sampled signals into uniformly sampled signals. For example, the EEG data can be resampled at 500 Hz.

[0073] ② Filtering. For example, filtering can be performed by notching. For example, EEG data, ECG data, and EMG data include 29 channels in total, of which EEG monitoring data includes 23 channels. All 29 channels can be filtered by 50Hz notch to reduce interference from AC signals, and the 23 EEG channels can be filtered by bandpass (such as using a frequency band of 0.1~70Hz) to reduce signal interference in non-EEG signal bands.

[0074] ③ Data enhancement. For example, the original lead, average lead, left hemisphere lead, and right hemisphere lead data of the EEG can be obtained, and all the lead data are spliced ​​to obtain enhanced EEG data.

[0075] ④ Eliminate noise data. Noise data may be caused by factors such as poor contact (such as poor contact between the biomedical monitoring equipment and the parts of the object being tested, poor contact between the equipment cables, etc.). Eliminating noise data can improve data quality and the accuracy of test results. For example, channel data with more noise can be eliminated.

[0076] ⑤ Numerical standardization processing. Numerical standardization processing can map different modalities and different types of data into the same appropriate numerical range for unified processing. For example, different types of original detection data such as EEG data, ECG data, and EMG data can be dimensioned first, and then normalized to an appropriate numerical range, such as by multiplying by corresponding coefficients for numerical mapping to complete numerical standardization processing.

[0077] ⑥ Time-frequency transformation. The original EEG and other detection data are usually time domain data, expressing the changes of electrical signals over time, and the information is relatively simple. Through time-frequency transformation, the original detection data can be converted to the frequency domain or the joint time-frequency domain to obtain frequency domain data or time-frequency data, which can provide the distribution information of the signal in the frequency domain, or the joint distribution information in the time domain and the frequency domain. The time-frequency transformation methods include but are not limited to short-time Fourier transform, wavelet transform, etc. The present disclosure does not limit which specific method is adopted.

[0078] According to actual needs and the characteristics of the original detection data, any one or more of the above preprocessing methods can be used, and different preprocessing methods can be used for different original detection data. For example, the original EEG data can be processed by filtering, data enhancement, and time-frequency transformation in sequence to obtain preprocessed EEG data.

[0079] Continue to refer Figure 2 In step S220, a first electrical signal feature and a first sleep feature are extracted from the biomedical detection data.

[0080] In the embodiments of the present disclosure, the electrical signal feature is a feature extracted based on the distribution law of the electrical signal data, and the sleep feature is a feature related to the sleep information of the subject. The first electrical signal feature and the first sleep feature are the electrical signal feature and the sleep feature extracted in the epileptic discharge detection task. The second electrical signal feature and the second sleep feature below are the electrical signal feature and the sleep feature extracted in the sleep state detection task.

[0081] Exemplarily, the biomedical detection data includes EEG data, such as preprocessed time-frequency EEG data. The distribution of EEG data can be counted, such as sampling or segmenting the EEG data according to a certain window, counting the maximum value, minimum value, average value, median, standard deviation and other index values ​​in each window, and using the statistical results as the first electrical signal feature. The sleep physiological wave in the EEG data is detected, and the characteristics of the sleep physiological wave are extracted, such as the position, width, peak height, etc. of the peak, or the data of the sleep physiological wave are counted to obtain the first sleep feature.

[0082] Continue to refer Figure 2 In step S230, it is determined whether the subject has epileptic discharge based on the first electrical signal characteristic and the first sleep characteristic.

[0083] For example, the first electrical signal feature and the first sleep feature can be input into a pre-trained classification model to obtain a classification result of whether the subject has an epileptic discharge, that is, an epileptic discharge detection result. Alternatively, the first electrical signal feature and the first sleep feature can be calculated according to a pre-set function or formula to obtain a probability value of the subject having an epileptic discharge, and the probability value can be used as an epileptic discharge detection result.

[0084] based on Figure 2 The method, in the task of epileptic discharge detection, in addition to extracting the first electrical signal feature, also extracts the first sleep feature, and determines whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature. The first sleep feature provides the sleep information of the subject, and by combining with the first electrical signal feature, it can help the computer to distinguish similar epileptic discharges from sleep physiological waves, thereby reducing the interference of sleep physiological waves on epileptic discharge detection, improving the accuracy and reliability of epileptic discharge detection results, and providing doctors with effective auxiliary diagnosis and treatment information. It also reduces the workload of manual post-processing and improves detection efficiency.

[0085] In one embodiment, the extracting of the first electrical signal feature and the first sleep feature from the biomedical detection data may include the following steps:

[0086] The biomedical detection data is input into a first feature network to obtain a first feature tensor; the first feature tensor includes a first electrical signal feature and a first sleep feature.

[0087] The first feature network and the second feature network share at least some parameters; the first feature network is a network used to extract features in an epileptic discharge detection task, and the second feature network is a network used to extract features in a sleep state detection task. The first feature network and the second feature network can be any type of neural network, for example, both can be EfficientNet (a convolutional neural network). In addition, the first feature network and the second feature network can both be pre-trained networks, or networks that have been trained (such as fine-tuning) on ​​a specific data set.

[0088] It should be understood that due to the guiding role of the task, the first feature network tends to extract features related to discharge, so the first electrical signal feature can be extracted. The second feature network tends to extract features related to sleep information. Since the first feature network and the second feature network share at least part of the parameters, the first feature network can also extract features related to sleep information, that is, the first sleep feature. The biomedical detection data is input into the first feature network, and the first feature tensor is output. The first feature tensor may include multiple dimensions, one part of which is the first electrical signal feature, and another part of which is the first sleep feature. Features can be extracted very efficiently and quickly through the first feature network.

[0089] In one embodiment, the above-mentioned determining whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature may include the following steps:

[0090] The first feature tensor is input into the first classification network to obtain a first classification result; the first classification result indicates whether the subject has epileptic discharge.

[0091] The first classification network is a network for classifying epileptic discharge detection results. It can be any type of neural network, such as a network formed by one or more fully connected layers. The first classification network can be a pre-trained network, or a network trained on a specific data set.

[0092] The first feature tensor is input into the first classification network, and after processing, the first classification result is output. The first classification result can be a binary classification result, indicating whether the subject has an epileptic discharge, such as the first classification result of 0 indicates that no epileptic discharge has occurred, and 1 indicates that an epileptic discharge has occurred. Alternatively, the first classification result can be a multi-classification result, indicating the degree or type of epileptic discharge in the subject, such as the first classification result of 0 indicates that no epileptic discharge has occurred, 1 indicates that an epileptic discharge has occurred and the degree is level 1, and 2 indicates that an epileptic discharge has occurred and the degree is level 2. The first feature tensor can be processed very efficiently and quickly through the first classification network to obtain an epileptic discharge detection result.

[0093] In one embodiment, reference Figure 4 As shown, the epileptic discharge detection method may further include the following steps S410 and S420:

[0094] Step S410, inputting the biomedical detection data into a second feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature;

[0095] Step S420, inputting the second feature tensor into the second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0096] in, Figure 4 The method shown is used to detect the sleep state of the subject. The sleep state detection can be combined with the epileptiform discharge detection to help doctors more comprehensively evaluate the health status of the subject. In addition, since the epileptiform discharge detection task and the sleep state detection task use the same biomedical detection data, there is no need to collect additional data, which is conducive to reducing data detection costs and improving data utilization.

[0097] The second feature network is a network used to extract features in the sleep state detection task. It should be understood that due to the guiding role of the task, the second feature network tends to extract features related to sleep information, so the second sleep feature can be extracted. The first feature network tends to extract features related to discharge. Since the first feature network and the second feature network share at least part of the parameters, the second feature network can also extract features related to discharge, that is, the second sleep feature. The biomedical detection data is input into the second feature network, and the second feature tensor is output. The second feature tensor may include multiple dimensions, one part of which is the second electrical signal feature, and the other part of which is the second sleep feature. Features can be extracted very efficiently and quickly through the second feature network.

[0098] The second classification network is a network used for classifying sleep state detection results. It can be any type of neural network, such as a network formed by one or more fully connected layers. The second classification network can be a pre-trained network, or a network trained on a specific data set.

[0099] The second feature tensor is input into the second classification network, and after processing, the second classification result is output. The second classification result can be a binary classification result, indicating whether the subject is in a sleeping state, such as the second classification result of 0 indicates that the subject is not in a sleeping state (ie, awake state), and 1 indicates that the subject is in a sleeping state. Alternatively, the second classification result can be a multi-classification result, indicating which sleep stage the subject is in, such as the second classification result of 0 indicates that the subject is not in a sleeping state (ie, awake state), 1 indicates that the subject is in the N1 (light sleep) stage, 2 indicates that the subject is in the N2 (light sleep) stage, 3 indicates that the subject is in the N3 (deep sleep) stage, and 4 indicates that the subject is in the REM (Rapid Eye Movement Sleep) stage.

[0100] Through the second feature network and the second classification network, the sleep state detection task can be performed very efficiently and quickly, and a more accurate sleep state detection result can be obtained.

[0101] In one embodiment, reference Figure 5 As shown, the epileptic discharge detection method may further include the following steps S510 to S550:

[0102] Step S510, inputting the sample biomedical detection data into the first feature network to obtain a first sample feature tensor; the first sample feature tensor includes a first sample electrical signal feature and a first sample sleep feature;

[0103] Step S520, inputting the first sample feature tensor into the first classification network to obtain a first sample classification result;

[0104] Step S530, inputting the biomedical detection data into a second feature network to obtain a second sample feature tensor; the second sample feature tensor includes a second sample electrical signal feature and a second sample sleep feature;

[0105] Step S540, inputting the second sample feature tensor into the second classification network to obtain a second sample classification result;

[0106] Step S550, updating the first feature network and the first classification network according to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label; the first label is the epileptic discharge label corresponding to the sample biomedical detection data, and the second label is the sleep state label corresponding to the sample biomedical detection data.

[0107] in, Figure 5 The method shown is used to train the first feature network and the first classification network. In the embodiments of the present disclosure, biomedical test data of a large number of patients can be collected as sample biomedical test data, and corresponding epileptic discharge detection results and sleep state detection results can be collected as labels to form a training data set. In the training data set, each set of data includes a sample biomedical test data, a corresponding first label and a second label. The first label can be an manually labeled epileptic discharge detection result, and the second label can be an manually labeled sleep state detection result. Of course, the first label or the second label may be missing in some data.

[0108] Figure 6 A schematic diagram of manually annotated data is shown. A multi-channel EEG can be collected by an EEG machine, the EEG data can be recorded in segments, and then the standard EEG is manually segmented to see if there is epileptic discharge, and the sleep state is marked, thereby obtaining sample biomedical detection data (such as EEG or segmented EEG), the corresponding first label and second label.

[0109] The sample biomedical detection data is input into the first feature network to obtain the first sample feature tensor, which is then input into the first classification network to obtain the first sample classification result. The biomedical detection data is input into the second feature network to obtain the second sample feature tensor, which is then input into the second classification network to obtain the second sample classification result. According to the difference between the first sample classification result and the first label, the first feature network and the first classification network are updated. And according to the difference between the second sample classification result and the second label, the first feature network and the first classification network are updated. Among them, the difference between the first sample classification result and the first label reflects the performance of the first feature network and the first classification network in the epileptic discharge detection task, and the greater the difference, the worse the performance. According to the difference between the first sample classification result and the first label, the update amount of the parameters in the first feature network and the first classification network can be calculated by back propagation, and the parameters are updated, so as to promote the first feature network and the first classification network to improve the performance in the epileptic discharge detection task. The difference between the second sample classification result and the second label reflects the performance of the first feature network and the first classification network in the sleep state detection task, especially the quality of the first sample sleep feature extracted by the first feature network. According to the difference between the second sample classification result and the second label, the update amount of the parameters in the first feature network and the first classification network can be calculated by back propagation, and the parameters are updated (mainly updating some parameters in the first feature network). This enables the first feature network and the first classification network to improve their performance in the sleep state detection task, especially to improve the quality of sleep feature extraction.

[0110] Based on the above method, multi-task joint training of epileptiform discharge detection task and sleep state detection task is realized. It not only promotes the performance of the first feature network and the first classification network in the epileptiform discharge detection task, but also promotes the first feature network to improve the extraction quality of sleep features, so that it can more accurately and fully learn the sleep information in the biomedical detection data, thereby further improving the first classification network's ability to distinguish similar epileptiform discharges and sleep physiological waves, and improving the accuracy and reliability of epileptiform discharge detection results.

[0111] In one embodiment, if a group of data contains sample biomedical test data and a first label but no second label, the sample biomedical test data can be input into a first feature network to obtain a first sample feature tensor, the first sample feature tensor can be input into a first classification network to obtain a first sample classification result, and the first feature network and the first classification network can be updated according to the difference between the first sample classification result and the first label. If a group of data contains sample biomedical test data and a second label but no first label, the sample biomedical test data can be input into a second feature network to obtain a second sample feature tensor, the second sample feature tensor can be input into a second classification network to obtain a second sample classification result, and the first feature network and the first classification network can be updated according to the difference between the second sample classification result and the second label.

[0112] In one embodiment, the epileptic discharge detection method may further include the following steps:

[0113] According to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label, the second feature network and the second classification network are updated.

[0114] Among them, the difference between the first sample classification result and the first label reflects the performance of the second feature network and the second classification network in the epileptic discharge detection task, and in particular, can reflect the quality of the second sample electrical signal feature extracted by the second feature network. According to the difference between the first sample classification result and the first label, the update amount of the parameters in the second feature network and the second classification network can be calculated by back propagation, and the parameters are updated (mainly updating some parameters in the second feature network), so that the second feature network and the second classification network can improve the performance in the epileptic discharge detection task, especially the second feature network can improve the extraction quality of the electrical signal feature. The difference between the second sample classification result and the second label reflects the performance of the second feature network and the second classification network in the sleep state detection task, and the greater the difference, the worse the performance. According to the difference between the second sample classification result and the second label, the update amount of the parameters in the second feature network and the second classification network can be calculated by back propagation, and the parameters are updated, so that the second feature network and the second classification network can improve the performance in the sleep state detection task. Through the training method of combining the two tasks, the second feature network and the second classification network are prompted to improve their performance in the sleep state detection task, and the second feature network is also prompted to improve the quality of electrical signal feature extraction, so that it can more accurately and fully learn the electrical signal information in the biomedical detection data, thereby further improving the second classification network's ability to distinguish similar epileptic discharges and sleep physiological waves, and improving the accuracy and reliability of sleep state detection results.

[0115] Figure 7 The implementation architecture diagram of the embodiment of the present disclosure is shown. The original test data of the test object, including the multi-channel EEG, is collected, and the EEG is segmented into 4-second segments to obtain EEGs of different time segments. Short-time Fourier transform is performed to obtain EEGs of different channels (including Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1 and other channels) in the time-frequency joint domain, that is, biomedical test data. The EEG is input into the first feature network. The first feature network includes Conv (convolutional layer, Figure 7 Conv3 in 3 means the convolution kernel size of the convolution layer is 3 3), MBConv (Mobile Inverted ResidualBottleneck Convolution, mobile inverted residual bottleneck convolution layer, Figure 7 6 of 6 MBConv4 represents 6 MBConv intermediate layers, whose scale is 4), F-MBConv (Fused-MBConv, fused mobile inverted residual bottleneck convolution layer), Pooling (pooling layer) and other intermediate layers. The first feature network outputs a first feature tensor, which includes a first electrical signal feature and a first sleep feature. The first feature tensor is input into a first classification network, which may include one or more fully connected layers, and the first feature tensor is subjected to epileptic discharge detection and classification to obtain the probability of epileptic discharge, such as the probability of detecting epileptic discharge is 0.965, and the probability of no discharge (i.e., no epileptic discharge) is 0.035. Since the probability of epileptic discharge is greater than the first preset probability value (which can be set according to experience, such as 0.7 or 0.8, etc.), the final detection result may be the presence of epileptic discharge. The electroencephalogram is input into the second feature network. The structure of the second feature network can be the same as that of the first feature network, and the parameters of the two are different, and some parameters can be shared. The second feature network outputs a second feature tensor, which includes a second electrical signal feature and a second sleep feature. The second feature tensor is input into the second classification network, which may include one or more fully connected layers, and the second feature tensor is subjected to epileptic sleep state detection and classification to obtain the probability of the sleep state, such as the probability of being in the sleep state is 0.956, and the probability of being in the awake state is 0.044. Since the probability of being in the sleep state is greater than the second preset probability value (which can be set according to experience, such as 0.7 or 0.8, etc.), the final detection result may be that the person is in the sleep state.

[0116] In one embodiment, the extracting of the first electrical signal feature and the first sleep feature from the biomedical detection data may include the following steps:

[0117] The biomedical test data is input into the third feature network to obtain basic features;

[0118] Inputting the basic features into the fourth feature network to obtain a first feature tensor; the first feature tensor includes a first electrical signal feature and a first sleep feature;

[0119] Among them, the third feature network is a network used to extract basic features in epileptic discharge detection tasks and sleep state detection tasks. The basic features can be preliminary features shared by electrical signal features and sleep features. The fourth feature network is a network used to further extract the first electrical signal features and the first sleep features based on the basic features in the epileptic discharge detection task. The third feature network and the fourth feature network can be any type of neural network, such as the third feature network is a convolutional neural network, and the fourth feature network is EfficientNet. It should be understood that the third feature network can be shared in the epileptic discharge detection task and the sleep state detection task, which is conducive to reducing training costs.

[0120] In the disclosed embodiment, the feature extraction process can be divided into two stages. In the first stage, the basic features are extracted through the third feature network, and in the second stage, the first feature tensor is extracted from the basic features through the fourth feature network, including the first electrical signal feature and the first sleep feature. This is conducive to extracting high-quality features and subsequently realizing accurate epileptic discharge detection.

[0121] In one embodiment, the epileptic discharge detection method may further include the following steps:

[0122] Inputting the basic features into the fifth feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature;

[0123] The second feature tensor is input into the second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0124] Among them, the fifth feature network is a network used to further extract the second electrical signal feature and the second sleep feature for the basic feature in the sleep state detection task. The fifth feature network can be any type of neural network, such as EfficientNet. It should be understood that in the sleep state detection task, the feature extraction process can also be divided into two stages. The first stage extracts the basic features through the third feature network, and the second stage extracts the second feature tensor from the basic features through the fifth feature network, including the second electrical signal feature and the second sleep feature. This is conducive to the extraction of high-quality features and the subsequent accurate sleep state detection.

[0125] Exemplary Devices

[0126] The following is a description of the epileptiform discharge detection device in the embodiment of the present disclosure. Figure 8 As shown, the epileptic discharge detection device 800 may include the following modules:

[0127] The data acquisition module 810 is configured to acquire biomedical detection data of the subject;

[0128] A feature extraction module 820 is configured to extract a first electrical signal feature and a first sleep feature from the biomedical detection data;

[0129] The classification module 830 is configured to determine whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature.

[0130] The above modules can be program modules or hardware modules.

[0131] In one embodiment, extracting a first electrical signal feature and a first sleep feature from biomedical detection data includes: inputting the biomedical detection data into a first feature network to obtain a first feature tensor; the first feature tensor includes the first electrical signal feature and the first sleep feature; wherein the first feature network and the second feature network share at least some parameters; the first feature network is a network for extracting features in an epileptic discharge detection task, and the second feature network is a network for extracting features in a sleep state detection task.

[0132] In one embodiment, determining whether an epileptic-like discharge occurs in a subject based on a first electrical signal feature and a first sleep feature includes: inputting a first feature tensor into a first classification network to obtain a first classification result; the first classification result indicates whether an epileptic-like discharge occurs in the subject.

[0133] In one embodiment, the device also includes a training module, which is configured to: input the sample biomedical detection data into a first feature network to obtain a first sample feature tensor; the first sample feature tensor includes a first sample electrical signal feature and a first sample sleep feature; input the first sample feature tensor into a first classification network to obtain a first sample classification result; input the biomedical detection data into a second feature network to obtain a second sample feature tensor; the second sample feature tensor includes a second sample electrical signal feature and a second sample sleep feature; input the second sample feature tensor into a second classification network to obtain a second sample classification result; update the first feature network and the first classification network according to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label; the first label is the epileptic discharge label corresponding to the sample biomedical detection data, and the second label is the sleep state label corresponding to the sample biomedical detection data.

[0134] In one embodiment, the training module is further configured to update the second feature network and the second classification network according to the difference between the first sample classification result and the first label and the difference between the second sample classification result and the second label.

[0135] In one embodiment, the feature extraction module 820 is further configured to: input the biomedical detection data into a second feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; the classification module is further configured to: input the second feature tensor into a second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0136] In one embodiment, extracting a first electrical signal feature and a first sleep feature from biomedical detection data includes: inputting the biomedical detection data into a third feature network to obtain basic features; inputting the basic features into a fourth feature network to obtain a first feature tensor; the first feature tensor includes the first electrical signal feature and the first sleep feature; wherein the third feature network is a network for extracting basic features in epileptic discharge detection tasks and sleep state detection tasks.

[0137] In one embodiment, the feature extraction module 820 is further configured to: input the basic features into the fifth feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; the classification module is further configured to: input the second feature tensor into the second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

[0138] In one embodiment, obtaining biomedical test data of a test subject includes: preprocessing the original test data of the test subject to obtain biomedical test data; the preprocessing includes one or more of the following processes: resampling, filtering, data enhancement, noise data removal, numerical normalization, and time-frequency transformation.

[0139] In addition, other specific details of the embodiments of the present disclosure have been described in detail in the embodiments of the above method and will not be repeated here.

[0140] Exemplary Program Products

[0141] The computer program product in the embodiment of the present disclosure is described below. The computer program product includes a computer program, and when the computer program is executed by a processor, the above method of the present disclosure is implemented.

[0142] In one embodiment, the computer program product may be a tangible product, such as a computer-readable storage medium storing a computer program. The readable storage medium may be based on electrical, magnetic, optical, electromagnetic, infrared, and other signals, including but not limited to: random access memory (RAM), read-only memory (ROM), magnetic tape, floppy disk, flash memory (Flash), mechanical hard disk (HDD), solid state drive (SSD), and the like. Exemplarily, the computer program product may be a non-volatile storage medium storing a computer program, such as a read-only memory, a NAND flash memory (Nand Flash), and the like.

[0143] In one embodiment, the computer program product may be an intangible product. For example, the computer program product may be a virtual digital product, such as an executable file or installation package containing a computer program.

[0144] The code of the computer program can be written in one or more programming languages. Programming languages ​​include, but are not limited to, C, Java, C++, etc. The program code can be executed entirely on the user computing device, or partially on the user computing device, or as a separate software package, or 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 can be connected to the user computing device through any type of network, such as a local area network (LAN), a wide area network (WAN), etc., or can be connected to an external computing device (e.g., an Internet connection provided by an operator).

[0145] Computer programs can be carried or transmitted through electrical, magnetic, optical, electromagnetic, infrared and other signals. Electronic devices can convert signals carrying computer programs into digital signals to run the computer programs. When a computer program is run on an electronic device, its code is used to enable the electronic device to execute (more specifically, the processor of the electronic device can execute) the method steps of various embodiments of the present disclosure, for example: step S210, obtaining biomedical detection data of the subject; step S220, extracting a first electrical signal feature and a first sleep feature from the biomedical detection data; step S230, determining whether the subject has an epileptic discharge based on the first electrical signal feature and the first sleep feature.

[0146] The above steps are performed by a computer program. In the task of detecting epileptic discharges, in addition to extracting the first electrical signal feature, the first sleep feature is also extracted, and whether the subject has epileptic discharges is determined based on the first electrical signal feature and the first sleep feature. The first sleep feature provides the sleep information of the subject, and by combining with the first electrical signal feature, it can help the computer distinguish similar epileptic discharges from sleep physiological waves, thereby reducing the interference of sleep physiological waves on epileptic discharge detection, improving the accuracy and reliability of epileptic discharge detection results, and providing doctors with effective auxiliary diagnosis and treatment information. It also reduces the workload of manual post-processing and improves detection efficiency.

[0147] Exemplary Electronic Devices

[0148] The electronic device in the embodiment of the present disclosure is described below. The electronic device may be Figure 1A or Figure 1B Any device in the present invention. The electronic device includes a processor and a memory, and the memory is used to store executable instructions of the processor. The processor is configured to execute the above method of the present invention by executing the executable instructions.

[0149] refer to Fig. 9 An electronic device according to an embodiment of the present disclosure is exemplified. Fig. 9The electronic device 900 shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0150] like Fig. 9 As shown, the electronic device 900 is presented in the form of a general computing device. The components of the electronic device 900 may include, but are not limited to: a processor 910, a memory 920, a bus 930 connecting different system components (including the memory 920 and the processor 910), an I / O (input / output) interface 940, and a network adapter 950.

[0151] The memory 920 stores program codes, which can be executed by the processor 910, so that the processor 910 executes the method steps of the embodiment of the present disclosure. For example, the processor 910 can execute the following steps: step S210, obtaining biomedical detection data of the subject; step S220, extracting the first electrical signal feature and the first sleep feature from the biomedical detection data; step S230, determining whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature.

[0152] The processor 910 performs the above steps. In the task of detecting epileptic discharges, in addition to extracting the first electrical signal feature, the first sleep feature is also extracted, and whether the subject has epileptic discharges is determined based on the first electrical signal feature and the first sleep feature. The first sleep feature provides the sleep information of the subject, and by combining with the first electrical signal feature, it can help the computer to distinguish similar epileptic discharges from sleep physiological waves, thereby reducing the interference of sleep physiological waves on epileptic discharge detection, improving the accuracy and reliability of epileptic discharge detection results, and providing doctors with effective auxiliary diagnosis and treatment information. It also reduces the workload of manual post-processing and improves detection efficiency.

[0153] The memory 920 may include a volatile memory, such as a random access memory (RAM) 921 and / or a cache unit 922, and may also include a non-volatile memory, such as a read-only memory (ROM) 923. The memory 920 may also include one or more program modules 924, such program modules 924 include but are not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or a certain combination may include the implementation of a network environment. For example, the program module 924 may include each module in the above-mentioned device.

[0154] The processor 910 may include processing units such as an AP (Application Processor), a modem processor, a GPU (Graphics Processing Unit), an ISP (Image Signal Processor), a controller, an encoder, a decoder, a DSP (Digital Signal Processor), a baseband processor and / or an NPU (Neural-Network Processing Unit).

[0155] The bus 930 is used to connect different parts of the electronic device 900 and may include a data bus, an address bus, a control bus, and the like.

[0156] The electronic device 900 may also communicate with one or more external devices 1000 (eg, a keyboard, a pointing device, a Bluetooth device, etc.), and such communication may be performed through the I / O interface 940 .

[0157] The electronic device 900 can also communicate with one or more networks through the network adapter 950. For example, the network adapter 950 can provide mobile communication solutions such as 3G / 4G / 5G, or provide wireless communication solutions such as wireless LAN, Bluetooth, near field communication, etc. The network adapter 950 can communicate with other modules of the electronic device 900 through the bus 930.

[0158] although Fig. 9 Not shown, other hardware and / or software modules may also be provided in the electronic device 900, including but not limited to: a display, a microcode, a device driver, a redundant processing unit, an external disk drive array, a RAID (Redundant Arrays of Independent Disks) system, a tape drive, and a data backup storage system.

[0159] It should be noted that, although several modules or submodules of the device are mentioned in the above detailed description, such division is merely exemplary and not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided into multiple units / modules to be embodied.

[0160] In addition, although the operations of the disclosed method are described in a specific order in the drawings, this does not require or imply that the operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step, and / or one step may be decomposed into multiple steps.

[0161] Although the spirit and principle of the present disclosure have been described with reference to several specific embodiments, it should be understood that the present disclosure is not limited to the disclosed specific embodiments, and the division of various aspects does not mean that the features in these aspects cannot be combined to benefit, and such division is only for the convenience of expression. The present disclosure is intended to cover various modifications and equivalent arrangements included in the spirit and scope of the attached claims.

Claims

1. A method for detecting epileptic discharges, characterized in that: The method comprises: Acquiring biomedical test data of the subject; the biomedical test data includes electroencephalogram data; extracting a first electrical signal feature and a first sleep feature from the biomedical detection data; Determining whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature; The step of extracting the first electrical signal feature and the first sleep feature from the biomedical detection data comprises: inputting the biomedical detection data into a first feature network to obtain a first feature tensor; the first feature tensor comprises the first electrical signal feature and the first sleep feature; The first feature network and the second feature network share at least some parameters; the first feature network is a network used to extract features in an epileptic discharge detection task, and the second feature network is a network used to extract features in a sleep state detection task; The determining whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature includes: inputting the first feature tensor into a first classification network to obtain a first classification result; the first classification result indicates whether the subject has epileptic discharge; The method also includes: inputting the sample biomedical detection data into the first feature network to obtain a first sample feature tensor; the first sample feature tensor includes a first sample electrical signal feature and a first sample sleep feature; inputting the first sample feature tensor into the first classification network to obtain a first sample classification result; inputting the biomedical detection data into the second feature network to obtain a second sample feature tensor; the second sample feature tensor includes a second sample electrical signal feature and a second sample sleep feature; inputting the second sample feature tensor into the second classification network to obtain a second sample classification result; updating the first feature network and the first classification network according to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label; the first label is an epileptic discharge label corresponding to the sample biomedical detection data, and the second label is a sleep state label corresponding to the sample biomedical detection data.

2. The method according to claim 1, characterized in that The method further comprises: The second feature network and the second classification network are updated according to the difference between the first sample classification result and the first label and the difference between the second sample classification result and the second label.

3. The method according to claim 1, characterized in that The method further comprises: Inputting the biomedical detection data into the second feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; The second feature tensor is input into a second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

4. The method according to claim 1, characterized in that: The extracting the first electrical signal feature and the first sleep feature from the biomedical detection data includes: Inputting the biomedical detection data into a third feature network to obtain basic features; Inputting the basic features into a fourth feature network to obtain a first feature tensor; the first feature tensor includes the first electrical signal feature and the first sleep feature; Among them, the third feature network is a network used to extract basic features in epileptic discharge detection tasks and sleep state detection tasks.

5. The method according to claim 4, characterized in that The method further comprises: Inputting the basic features into a fifth feature network to obtain a second feature tensor; the second feature tensor includes a second electrical signal feature and a second sleep feature; The second feature tensor is input into a second classification network to obtain a second classification result; the second classification result represents the sleep state of the subject.

6. The method according to claim 1, characterized in that The step of obtaining biomedical test data of the tested object includes: Preprocessing the original detection data of the detected object to obtain the biomedical detection data; The preprocessing includes one or more of the following processes: resampling, filtering, data enhancement, noise data removal, numerical standardization, and time-frequency transformation.

7. An epileptic discharge detection device, characterized in that: The device comprises: A data acquisition module is configured to acquire biomedical detection data of the subject; the biomedical detection data includes EEG data; a feature extraction module, configured to extract a first electrical signal feature and a first sleep feature from the biomedical detection data; a classification module, configured to determine whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature; The step of extracting the first electrical signal feature and the first sleep feature from the biomedical detection data comprises: inputting the biomedical detection data into a first feature network to obtain a first feature tensor; the first feature tensor comprises the first electrical signal feature and the first sleep feature; The first feature network and the second feature network share at least some parameters; the first feature network is a network used to extract features in an epileptic discharge detection task, and the second feature network is a network used to extract features in a sleep state detection task; The determining whether the subject has epileptic discharge according to the first electrical signal feature and the first sleep feature includes: inputting the first feature tensor into a first classification network to obtain a first classification result; the first classification result indicates whether the subject has epileptic discharge; The device also includes a training module, which is configured to: input the sample biomedical detection data into the first feature network to obtain a first sample feature tensor; the first sample feature tensor includes a first sample electrical signal feature and a first sample sleep feature; input the first sample feature tensor into the first classification network to obtain a first sample classification result; input the biomedical detection data into the second feature network to obtain a second sample feature tensor; the second sample feature tensor includes a second sample electrical signal feature and a second sample sleep feature; input the second sample feature tensor into the second classification network to obtain a second sample classification result; update the first feature network and the first classification network according to the difference between the first sample classification result and the first label, and the difference between the second sample classification result and the second label; the first label is the epileptic discharge label corresponding to the sample biomedical detection data, and the second label is the sleep state label corresponding to the sample biomedical detection data.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 6 by executing the executable instructions.

Images