Abnormal discharge detection method, device, program product and electronic equipment

Abstract

The embodiments of the present disclosure relate to abnormal discharge detection methods, devices, program products and electronic devices, and to the field of artificial intelligence technology. The above method includes: obtaining a biomedical signal to be detected from a subject, and determining a target category of the biomedical signal to be detected; determining a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models according to the target category; processing the biomedical signal to be detected based on the target abnormal discharge detection model, and obtaining an abnormal discharge detection result of the subject according to the processing result. The present disclosure improves the efficiency of abnormal discharge detection and recognition and the accuracy of automatic recognition of abnormal discharge detection results.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, and in particular to an abnormal discharge detection method, an abnormal 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] Electroencephalogram (EEG) is an important basis for judging whether a patient's brain function is abnormal, and is of great value for clinical diagnosis and treatment, and assessment of the severity of the disease. The rapid development of artificial intelligence technology has made it possible to realize automatic EEG analysis, especially neural network technology, which has shown good application potential in detecting abnormal EEG discharges. Summary of the invention

[0004] However, the detection accuracy and reliability of the automatic EEG abnormality detection methods in the related art are insufficient, making it difficult to achieve truly independent EEG signal abnormality detection.

[0005] Therefore, an abnormal discharge detection method is highly needed to improve the accuracy and reliability of automatic detection of EEG abnormalities.

[0006] In this context, embodiments of the present invention are intended to provide an abnormal discharge detection method, an abnormal discharge detection device, a computer-readable storage medium, a computer program product, and an electronic device.

[0007] According to a first aspect of an embodiment of the present disclosure, there is provided an abnormal discharge detection method, comprising: obtaining a biomedical signal to be detected of a test subject, and determining a target category of the biomedical signal to be detected; according to the target category, determining a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models; based on the target abnormal discharge detection model, processing the biomedical signal to be detected, and obtaining an abnormal discharge detection result of the test subject according to the processing result.

[0008] Optionally, obtaining the biomedical signal to be detected of the subject includes: obtaining the collected biomedical signal of the subject; obtaining the biomedical signal of the subject in segments according to preset time lengths, and determining at least one biomedical signal to be detected of the subject.

[0009] Optionally, the processing of the biomedical signal to be detected based on the target abnormal discharge detection model and obtaining the abnormal discharge detection result of the object to be detected according to the processing result includes: processing the biomedical signal to be detected based on the target abnormal discharge detection model, and obtaining the abnormal discharge detection result of the biomedical signal to be detected according to the processing result; marking the biomedical signal to be detected when the abnormal discharge detection result of the biomedical signal to be detected is abnormal discharge; and obtaining the abnormal discharge detection result of the object to be detected according to the marked biomedical signal to be detected.

[0010] Optionally, when the target category is the first category, the target abnormal discharge detection model is a first abnormal discharge detection model; based on the target abnormal discharge detection model, the biomedical signal to be detected is processed, and the abnormal discharge detection result of the object under test is obtained according to the processing result, including: inputting the biomedical signal to be detected into the first multi-head attention sub-model in the first abnormal discharge detection model, and obtaining the first feature of the biomedical signal to be detected according to the output of the first multi-head attention sub-model; inputting the first feature into the first deep convolutional neural network sub-model in the first abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the output of the first deep convolutional neural network sub-model.

[0011] Optionally, when the target category is the second category, the target abnormal discharge detection model is a second abnormal discharge detection model, and the second abnormal discharge detection model includes a second deep convolutional neural network sub-model; based on the target abnormal discharge detection model, the biomedical signal to be detected is processed, and the abnormal discharge detection result of the object under test is obtained according to the processing result, including: transforming the form of the biomedical signal to be detected to obtain image-like biomedical signal data corresponding to the biomedical signal to be detected; inputting the image-like biomedical signal data into the second deep convolutional neural network sub-model, and obtaining the abnormal discharge detection result of the object under test according to the output of the second deep convolutional neural network sub-model.

[0012] Optionally, the biomedical signal to be detected includes a first matrix, which corresponds to a first matrix dimension. The biomedical signal to be detected of the object under test is transformed in form to obtain image-like biomedical signal data corresponding to the biomedical signal to be detected; the first matrix is ​​deformed to obtain a second matrix, which corresponds to a second matrix dimension, and the difference between the number of rows and the number of columns of the second matrix dimension is less than a preset value; and the image-like biomedical signal data is determined according to the second matrix.

[0013] Optionally, any of the deep convolutional neural network sub-models includes any one of a deep dilated convolutional network sub-model and a multi-layer convolutional neural network sub-model.

[0014] According to a second aspect of an embodiment of the present disclosure, an abnormal discharge detection device is provided, comprising: a target category determination module, configured to obtain a biomedical signal to be detected of a test subject, and determine a target category of the biomedical signal to be detected; a model selection module, configured to determine a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models according to the target category; and an abnormal detection module, configured to process the biomedical signal to be detected based on the target abnormal discharge detection model, and obtain an abnormal discharge detection result of the test subject according to the processing result.

[0015] According to a third aspect of the present disclosure, there is provided a computer program product comprising instructions, which, when executed on a computer, enables the computer to execute the steps of the abnormal discharge detection method according to the first aspect.

[0016] According to a fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the abnormal discharge detection method as described in the first aspect of the above embodiment is implemented.

[0017] According to the fifth aspect of the embodiment of the present disclosure, an electronic device is provided, comprising: a processor; and a storage device for storing one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the abnormal discharge detection method as described in the first aspect of the above embodiment.

[0018] According to the abnormal discharge detection method, abnormal discharge detection device, computer-readable storage medium, computer program product and electronic device of the embodiment of the present disclosure, by classifying the biomedical signal, determining the appropriate target abnormal discharge detection model according to the category of the biomedical signal, and based on the target abnormal discharge detection model, the abnormal discharge of the biomedical signal is identified. On the one hand, according to the abnormal discharge detection model, the present disclosure can realize the automatic detection and identification of abnormal discharge of the biomedical signal, and improve the efficiency of abnormal discharge detection; on the other hand, according to the category of the biomedical signal, the present disclosure determines the suspected abnormal biomedical signal and its suspected abnormal category, selects the target abnormal discharge detection model corresponding to the category of the suspected abnormal biomedical signal, and performs targeted abnormal identification processing on the different categories of suspected abnormal biomedical signals again, so as to improve the accuracy of abnormal discharge detection of the biomedical signal; on the other hand, according to the abnormal discharge detection result obtained by the method of the present disclosure, it can provide doctors with more accurate reference information, so that doctors do not need to observe all the data in detail, but only roughly browse some key information, and then combine the abnormal discharge detection result, so as to quickly and accurately determine the diagnosis result of the patient, saving the workload of doctors and facilitating wide application in clinical practice. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the detailed description below with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0020] Figure 1 A schematic diagram showing an exemplary system architecture to which the embodiments of the present disclosure may be applied;

[0021] Figure 2 A schematic flow chart showing an abnormal discharge detection method in an exemplary embodiment of the present disclosure;

[0022] Figure 3 A schematic flow chart showing a method for acquiring a biomedical signal to be detected in an exemplary embodiment of the present disclosure;

[0023] Figure 4 A schematic flow chart showing a method for obtaining abnormal discharge detection results of a tested object in an exemplary embodiment of the present disclosure;

[0024] Figure 5 A flow chart showing a method for determining an abnormal discharge detection result of a first category of a biomedical signal to be detected in an exemplary embodiment of the present disclosure;

[0025] Figure 6A schematic structural diagram of a first abnormal discharge detection model in an exemplary embodiment of the present disclosure is shown;

[0026] Figure 7 A flow chart showing a method for determining an abnormal discharge detection result of a second category of a biomedical signal to be detected in an exemplary embodiment of the present disclosure;

[0027] Figure 8 A schematic diagram showing the composition of an abnormal discharge detection device in an exemplary embodiment of the present disclosure;

[0028] Fig. 9 A schematic structural diagram of an electronic device in an exemplary embodiment of the present disclosure is shown.

[0029] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0030] The principles and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way. On the contrary, these embodiments are provided to make the present invention more thorough and complete, and to fully convey the scope of the present invention to those skilled in the art.

[0031] Those skilled in the art will appreciate that the embodiments of the present invention may be implemented as a system, device, apparatus, medium, method or computer program product. Therefore, the present invention may be implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0032] According to an embodiment of the present invention, an abnormal discharge detection method, an abnormal discharge detection device, a computer program product, and an electronic device are provided.

[0033] In this document, any number of elements in the drawings is used for illustration rather than limitation, and any naming is used only for distinction and does not have any limiting meaning.

[0034] The principle and spirit of the present invention are described in detail below with reference to several representative embodiments of the present invention. SUMMARY OF THE INVENTION

[0036] The inventors of the present disclosure have found that the abnormal discharge detection method in the related art has the problem of low accuracy and reliability in detecting abnormal discharge of biomedical signals such as EEG.

[0037] In view of the above content, the basic idea of ​​the present disclosure is to provide an abnormal discharge detection method, device, program product and electronic device, by classifying biomedical signals, determining a suitable target abnormal discharge detection model according to the category of the biomedical signals, and identifying abnormal discharges of the biomedical signals based on the target abnormal discharge detection model. On the one hand, the present disclosure can realize the automatic detection and identification of abnormal discharges of biomedical signals according to the abnormal discharge detection model, and improve the efficiency of abnormal discharge detection; on the other hand, the present disclosure selects the target abnormal discharge detection model corresponding to the category of the biomedical signal according to the category of the biomedical signal, and performs targeted processing on biomedical signals of different categories to improve the accuracy of abnormal discharge detection of biomedical signals; on the other hand, the abnormal discharge detection result obtained according to the method of the present disclosure can provide doctors with more accurate reference information, so that doctors do not need to observe all the data in detail, but only roughly browse some key information, and then combine the abnormal discharge detection results to quickly and accurately determine the diagnosis results of the patient, saving the workload of doctors and facilitating wide application in clinical practice.

[0038] After introducing the basic principles of the present invention, various non-limiting embodiments of the present invention are described in detail below.

[0039] Application Scenario Overview

[0040] It should be noted that the following application scenarios are only shown to facilitate understanding of the spirit and principle of the present invention, and the embodiments of the present invention are not limited in this respect. On the contrary, the embodiments of the present invention can be applied to any applicable scenario.

[0041] The embodiments of the present disclosure can be applied to the abnormal discharge detection scenario of the human brain. For example, when the patient feels discomfort in the head, such as a headache, the brain electrical signals of the patient can be collected, and then the collected brain electrical signals of the patient can be classified and identified. According to the target category of the determined brain electrical signals of the patient, a target abnormal discharge detection model corresponding to the target category is determined from multiple pre-trained abnormal discharge detections, and then the collected brain electrical signals are input into the target abnormal discharge detection model, and the abnormal discharge detection result of the patient is obtained according to the output of the target abnormal discharge detection model. The abnormal discharge detection result of the patient is fed back to the doctor to assist the doctor in making a quick and accurate clinical diagnosis of the patient based on the abnormal discharge detection result combined with other key indicator data, thereby reducing the doctor's workload and improving the doctor's work efficiency.

[0042] Exemplary System Architecture

[0043] First, refer to Figure 1The system architecture of an exemplary application environment of the present disclosure is described.

[0044] like Figure 1 As shown, the system architecture 100 may include a terminal device 110 and a server 120. The terminal device 110 may be a terminal device such as a smart phone, a tablet computer, a desktop computer, a laptop computer, a smart wearable device, etc. The server 120 generally refers to a background system that provides services related to the abnormal discharge detection method in this exemplary embodiment, and may be a server or a cluster formed by multiple servers. The terminal device 110 and the server 120 may be connected via a wired or wireless communication link to exchange data.

[0045] In an exemplary embodiment, the above-mentioned abnormal discharge detection method can be performed by the server 120. Accordingly, the abnormal discharge detection device can be set in the server 120 to realize the corresponding module functions. For example, the biomedical signal of the subject is collected by the electroencephalograph, and the collected biomedical signal is sent to the server 120. The server 120 performs segmentation processing on the received biomedical signal to obtain the biomedical signal to be detected, and then determines the target category of the biomedical signal to be detected by a pre-trained classification model, and then determines the target abnormal discharge detection model according to the target category, and inputs the biomedical signal to be detected into the target abnormal discharge detection model, and determines the abnormal discharge detection result of the subject according to the output of the target abnormal discharge detection model. After the server 120 determines the abnormal discharge detection result, it can also send the abnormal discharge detection result to the client where the relevant personnel are located, such as the doctor's client.

[0046] In another exemplary embodiment, the abnormal discharge detection method described above may also be performed by the terminal device 110. Accordingly, the abnormal discharge detection device may be provided in the terminal device 110 to implement the corresponding module functions. For example, the user uploads the collected biomedical signal of the subject to the terminal device 110, and the terminal device 110 determines the target category of the biomedical signal uploaded by the user according to the classification model pre-configured in the terminal device 110, and determines the target abnormal discharge detection model from the multiple pre-trained abnormal discharge detection models configured in the terminal device 110 according to the target category, and then inputs the biomedical signal into the target abnormal discharge detection model, determines the abnormal discharge detection result according to the output of the target abnormal discharge detection model, and displays the determined abnormal discharge detection result in the graphical user interface of the terminal device 110.

[0047] It should be understood that Figure 1The number of terminal devices and servers in the example is only illustrative. Any number of terminal devices and servers may be provided as required. For example, the server 120 may be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms.

[0048] It is easy for those skilled in the art to understand that the above application scenarios are only used as examples and are not limited to the present exemplary embodiment.

[0049] Exemplary Methods

[0050] Figure 2 A flow chart showing an abnormal discharge detection method in an exemplary embodiment of the present disclosure is shown. Figure 2 , the method comprising:

[0051] Step S210, obtaining a biomedical signal to be detected from the subject, and determining a target category of the biomedical signal to be detected;

[0052] Step S220, according to the target category, determining a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models;

[0053] Step S230 , processing the biomedical signal to be detected based on the target abnormal discharge detection model, and obtaining an abnormal discharge detection result of the detected object according to the processing result.

[0054] Next, the specific implementation method of "step S210, obtaining the biomedical signal to be detected of the subject, and determining the target category of the biomedical signal to be detected" is first described in detail.

[0055] In an exemplary embodiment, the biomedical signal is an active signal spontaneously generated by the physiological process of a living body, which can more accurately reflect the physiological state or physical state of a person. Therefore, the biomedical signal can be used to determine whether the living body is abnormal. The living body in the present disclosure may include any living body, such as a human body, an animal body, etc., and this exemplary embodiment does not specifically limit this.

[0056] Exemplarily, the above-mentioned biomedical signals may include any one or more of the following: physiological signals such as electrocardiogram signals, electroencephalogram signals, electromyography signals, electrooculography signals and electrogastric signals, or may also include non-electrical signals such as body temperature, blood pressure, pulse, respiration, etc., which can be set according to actual conditions, and the present disclosure does not make any special limitations on this.

[0057] Among them, the ECG signal can be an electrocardiogram collected by a multi-channel electrocardiogram machine. An electrocardiogram (ECG) refers to a 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.

[0058] EEG signals can be electroencephalograms collected by multi-channel electroencephalogram machines. Electroencephalograms (EEGs) are graphs obtained by amplifying and recording the spontaneous bioelectric potential of the cerebral cortex from the scalp through sophisticated instruments. They are spontaneous and rhythmic electrical activities of brain cell groups 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 EEGs.

[0059] Electromyography (EMG) is the temporal and spatial superposition of action potentials of motor units in numerous muscle fibers, which can be obtained by sticking electromyography sensors on the skin.

[0060] 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.

[0061] Electrogastrogram (EGG) is an electrical signal generated by the contraction of stomach muscles, which can be collected on the surface of the abdominal skin using electrodes.

[0062] In an optional embodiment, the subjects may include patients who seek medical treatment for epilepsy or epilepsy-like symptoms, and the subjects may also include people who seek medical treatment for other brain discomfort, such as people with brain pain, etc. This exemplary embodiment does not specifically limit this.

[0063] For example, Figure 3 A schematic diagram of a method for acquiring a biomedical signal to be detected in an exemplary embodiment of the present disclosure is shown. Figure 3 , the method may include steps S310 to S320. Wherein:

[0064] In step S310, the collected biomedical signals of the subject are acquired.

[0065] For example, for each subject, a long-term biomedical signal can be collected, such as a 10-minute biomedical signal. Since abnormal discharge is not continuous, the biomedical signal of the subject can be processed in segments to obtain the biomedical signal to be detected, and whether the subject has abnormal discharge can be determined based on the biomedical signal to be detected, thereby improving the efficiency and accuracy of abnormal discharge detection.

[0066] In step S320, biomedical signals of the subject are acquired in segments according to preset time lengths, and at least one biomedical signal to be detected of the subject is determined.

[0067] Taking the preset time length of 4 seconds as an example, the biomedical signal collected in step S210 can be segmented into 4-second units, and each 4-second segmented signal is used as a biomedical signal to be detected. In this way, 15 biomedical signals to be detected can be obtained in 1 minute, and 150 biomedical signals to be detected can be obtained in 10 minutes.

[0068] In another exemplary embodiment, the biomedical signal collected by the biomedical signal device can also be directly used as the biomedical signal to be detected without segmenting it. For example, the biomedical signal for 1 minute can be directly collected as needed and the biomedical signal collected for 1 minute can be used as the biomedical signal to be detected. This exemplary embodiment does not make any special limitation on this.

[0069] For example, the biomedical signal acquisition equipment directly acquires the biomedical signal digital sequence. For example, a multi-channel electrocardiograph, a multi-channel electroencephalogram, or a multi-channel electromyogram can be used to acquire the corresponding biomedical signal at a preset acquisition frequency, such as 500 Hz, and the obtained biomedical signal numerical sequence is obtained.

[0070] Taking EEG signals as an example, the EEG signals collected by an EEG machine can be a sequence of EEG signal values ​​as shown in the following formula (1).

[0071] x_t=(x_t1,x_t2,x_t3,x_t4,…,x_tn) (1)

[0072] In formula (1), x_t is the EEG signal numerical sequence. x_t1, x_t2, x_t3, x_t4, …, x_tn represent the signal values ​​of the EEG signal at different sampling times. For example, x_t1 represents the signal value at the first sampling time in the EEG signal numerical sequence x_t, and x_tn represents the signal value at the nth sampling time in the EEG signal numerical sequence x_t.

[0073] It can be seen from formula (1) that the original biomedical signal collected by the biomedical signal acquisition device can be regarded as a one-dimensional or multi-dimensional long matrix. For example, if the number of channels of a multi-channel EEG signal is 29 and the signal is sampled 2000 times in 4 seconds, the biomedical signal to be detected is a 29-by-2000-dimensional long matrix. The abnormal discharge detection model is usually a neural network model. The neural network model, especially the convolutional neural network, has a natural processing advantage for image signals. When processing image signals, it can extract deeper feature expressions of image signals. Most image signals are multi-row and multi-column matrices with basically the same or similar number of rows and columns, such as 256 by 256, 512 by 512, etc. In other words, the original biomedical signal is similar to a rectangular input, while the image signal is similar to a square input. The two are quite different in form, which makes it impossible for neural networks, especially convolutional neural networks, to fully utilize the advantages of feature extraction in abnormal discharge detection.

[0074] That is to say, only when the form of the input biomedical signal of the abnormal discharge detection model is similar to that of the image signal, can the advantages of neural networks, especially convolutional neural networks, in feature extraction be fully applied to the field of abnormal discharge detection, extract deeper biomedical signal features, and improve the accuracy of biomedical signal detection.

[0075] Biomedical signals are related to time. As mentioned earlier, they are time series. When they are converted into image-based signals, the time information will be lost. For some abnormal discharge signals that rely heavily on time information, such as the phenomenon of continuous abnormal discharge in some objects, after converting them into image-based signals, although other deep-level features are extracted, the time information is lost, resulting in a decrease in the accuracy of the detection results when using convolutional neural networks for abnormality detection.

[0076] However, for some abnormal discharge signals that have low dependence on time information, such as the sudden and short-lived abnormal discharges of some objects, which have a low degree of dependence on time, after converting them into image-based signals, although the time information is lost, the impact on them is not significant. On the contrary, due to the processing advantages of convolutional neural networks for image-based signals, deeper biomedical signal features are extracted, thereby improving the accuracy of detection.

[0077] Based on this, in the present disclosure, biomedical signals can be classified, and suspected abnormal biomedical signals and suspected abnormal categories can be determined based on the classification results. Then, based on the characteristics of the suspected abnormal categories of the suspected abnormal biomedical signals, a suitable abnormal discharge detection method can be selected to confirm the abnormality of the suspected abnormal biomedical signals, thereby improving the accuracy and reliability of abnormal discharge detection.

[0078] In an exemplary embodiment, a classification model for biomedical signals can be pre-trained, and the biomedical signal to be detected can be input into the classification model. The target category of the biomedical signal to be detected can be determined based on the output of the classification model. Thus, based on the target category of the biomedical signal to be detected, it can be determined whether the biomedical signal to be detected is a suspected abnormal biomedical signal and the category of the suspected abnormality.

[0079] In an exemplary embodiment, the categories of biomedical signals in the present disclosure may include a first category, a second category, and a third category. The first category represents suspected abnormal biomedical signals that are dependent on time information or have a high degree of dependence, the second category represents suspected abnormal biomedical signals that are not dependent on time information or have a low degree of dependence, and the third category represents normal biomedical signals.

[0080] For example, the first data set, the second data set, and the third data set can be determined by manual labeling, wherein the first data set corresponds to the first category mentioned above, that is, the biomedical signals in the first data set are all of the first category, the second data set corresponds to the second category mentioned above, and the third data set corresponds to the third category mentioned above. For example, the categories of biomedical signals are manually labeled based on the discharge characteristics of existing biomedical signals. For example, for biomedical signals with continuous abnormal discharge, their categories are marked as the first category, for biomedical signals with short-term abnormal discharge, their categories are marked as the second category, and for biomedical signals without abnormal discharge, they are marked as the third category, thereby obtaining a training data set for the classification model.

[0081] It should be noted that the specific training method of the classification model and the loss function in the training process can refer to the relevant content of the existing machine learning model that can be used for classification, and will not be repeated here.

[0082] The classification model may include any machine learning model capable of achieving classification, such as a support vector machine, a decision tree model, etc., and this exemplary embodiment does not impose any special limitation on this.

[0083] After the classification model is trained, the classification model is mainly used to classify the biomedical signals to be detected. As mentioned above, the classification categories may include the first category, the second category and the normal category, wherein the first category and the second category both belong to the abnormal category, but the two belong to different abnormal situations. The classification model can quickly determine the biomedical signals similar to normal biomedical signals or abnormal biomedical signals, thereby quickly determining the suspected abnormal biomedical signals and the suspected abnormal categories corresponding to the suspected abnormal biomedical signals. If a biomedical signal to be detected is in the first category, it is a suspected abnormal biomedical signal and its suspected abnormal category is the above-mentioned first category. If a biomedical signal to be detected is in the second category, it is a suspected abnormal biomedical signal and its suspected abnormal category is the above-mentioned second category. If a biomedical signal to be detected is in the third category, it is a normal biomedical signal.

[0084] In an exemplary embodiment, the classification model can be a simple machine learning model, the main purpose of which is to perform a quick preliminary classification of biomedical signals, and determine the suspected abnormal biomedical signals and the categories of the suspected abnormalities according to the classification results. In subsequent steps, for different categories of suspected abnormalities, the abnormal discharge detection model corresponding to the category is used to perform deeper feature extraction to further confirm whether there is indeed an abnormality, thereby improving the accuracy of abnormality detection of biomedical signals.

[0085] The specific implementation of "step S220, determining a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models according to the target category" is described in detail below.

[0086] For example, multiple pre-trained abnormal discharge detection models can be obtained based on different training data. For example, supervised learning training can be performed on a machine learning model using the first data set and the third data set to obtain a first abnormal discharge detection model; supervised learning training can be performed on another machine learning model using the second data set and the third data set to obtain a second abnormal discharge detection model.

[0087] Among them, when training the abnormal discharge detection model, the labels of the biomedical signals in the first data set, the second data set and the third data set are all abnormal discharge labels, that is, the abnormal discharge labels are used to indicate whether the biomedical signals have abnormal discharge. If the biomedical signals in the first data set, the second data set and the third data set are all biomedical signals in the existing data set, then they will have their own labels of whether they have abnormal discharge. If they are not biomedical signals in the existing data set, the abnormal discharge labels corresponding to the biomedical signals in the first data set, the second data set and the third data set can be marked manually, such as by a doctor reading the image.

[0088] It should be noted that the specific implementation methods of training the first abnormal discharge detection model based on the biomedical signals in the first data set and the third data set and their corresponding abnormal discharge labels, and training the second abnormal discharge detection model based on the biomedical signals in the second data set and the third data set and their corresponding abnormal discharge labels can refer to the training methods of existing machine learning models, which will not be repeated here.

[0089] For example, the above-mentioned first abnormal discharge detection model is trained based on the first data set and the third data set, and the category label of the biomedical signal in the first data set is the first category. The above-mentioned second abnormal discharge detection model is trained based on the second data set and the third data set, and the category label of the biomedical signal in the second data set is the second category. Therefore, the first category can be pre-associated with the first abnormal discharge detection model, and the second category can be associated with the second abnormal discharge detection model, so as to obtain a mapping relationship table between the biomedical signal category and the pre-trained abnormal discharge detection model.

[0090] According to the mapping relationship table, it can be determined that the target abnormal discharge detection model of the first category of biomedical signals is the above-mentioned first abnormal discharge detection model, and the target abnormal discharge detection model of the second category of biomedical signals is the above-mentioned second abnormal discharge detection model. In other words, when the target category is the first category, according to the mapping relationship table, it can be determined that the target abnormal discharge detection model is the above-mentioned first abnormal discharge detection model, and when the target category is the second category, according to the mapping relationship table, it can be found that the target abnormal discharge detection model is the above-mentioned second abnormal discharge detection model.

[0091] In an exemplary embodiment, the abnormal discharge detection model in the present disclosure is determined according to a convolutional neural network. For example, the first abnormal discharge detection model corresponding to the first category is a first convolutional neural network, and the second abnormal discharge detection model corresponding to the second category is a second convolutional neural network.

[0092] In another exemplary embodiment, since the first category of biomedical signals has a strong dependence on time information, the time information features of the first category of biomedical signals can be extracted through a multi-head attention model. Based on this, the first abnormal discharge detection model can be composed of a first multi-head attention sub-model and a first deep convolutional neural network sub-model, and the second abnormal discharge detection model can include a second deep convolutional neural network. Since the second category of biomedical signals has a low dependence on time information, the second abnormal discharge detection model can be composed of only one second deep convolutional neural network sub-model.

[0093] In an exemplary embodiment, any deep convolutional neural network submodel includes any one of a deep dilated convolutional network submodel and a multi-layer convolutional neural network submodel. That is, both the first deep neural network submodel and the second deep convolutional neural network submodel can include any one of a deep dilated convolutional network submodel and a multi-layer convolutional neural network submodel. Of course, any deep convolutional neural network submodel can also include other convolutional network types, which is not particularly limited in this exemplary embodiment.

[0094] The specific implementation of "step S230, processing the biomedical signal to be detected based on the target abnormal discharge detection model, and obtaining the abnormal discharge detection result of the detected object according to the processing result" is described in detail below.

[0095] For example, Figure 4 A flow chart showing a method for obtaining abnormal discharge detection results of a tested object in an exemplary embodiment of the present disclosure is shown. Figure 4 , the method may include steps S410 to S430. Wherein:

[0096] In step S410, the biomedical signal to be detected is processed based on the target abnormal discharge detection model, and an abnormal discharge detection result of the biomedical signal to be detected is obtained according to the processing result.

[0097] For example, the biomedical signal to be detected can be input into the target abnormal discharge detection model to process the biomedical signal to be detected, and an abnormal discharge detection result of whether the biomedical signal to be detected has abnormal discharge can be obtained according to the output of the target abnormal discharge detection model.

[0098] In step S420, when the abnormal discharge detection result of the biomedical signal to be detected is abnormal discharge, the biomedical signal to be detected is marked.

[0099] For example, when there is abnormal discharge in the biomedical signal to be detected, an abnormal mark, such as the first mark, can be added to the biomedical signal to be detected. Since the biomedical signal to be detected is acquired in segments, the abnormal mark can be used to locate the time period in which the abnormal discharge occurs in the subject, which is convenient for the doctor to make subsequent diagnosis and confirmation.

[0100] Exemplarily, when the abnormal discharge detection result of the biomedical signal to be detected is normal discharge, no mark may be added to it, or a normal mark, such as the second mark, may be added to it.

[0101] In step S430, the abnormal discharge detection result of the detected object is obtained according to the marked biomedical signal to be detected.

[0102] Exemplarily, the time periods corresponding to the marked biomedical signals to be detected can be summarized to obtain the time periods in which the subject has abnormal discharges, and then the time periods in which the subject has abnormal discharges are used as abnormal discharge detection results and fed back to the client where the doctor is located. The doctor can quickly and accurately locate the time period in which abnormal discharges may exist based on the abnormal discharge detection results, and study the biomedical signal images in the time period in which abnormal discharges may exist to confirm whether abnormal discharges are really occurring. In this way, the doctor does not need to study the entire biomedical signal image as a whole, but only needs to confirm the abnormality of the biomedical signal images in the abnormal time period summarized in the detection results, thereby improving the doctor's work efficiency.

[0103] For example, Figure 5 A flow chart showing a method for determining an abnormal discharge detection result of a first category of a biomedical signal to be detected in an exemplary embodiment of the present disclosure is shown. Figure 5 , the method may include steps S510 to S520.

[0104] In step S510, the biomedical signal to be detected is input into a first multi-head attention sub-model in a first abnormal discharge detection model, and a first feature of the biomedical signal to be detected is obtained according to an output of the first multi-head attention sub-model.

[0105] For example, when the biomedical signal to be detected is of the first category, the target abnormal discharge detection model is the first abnormal discharge detection model mentioned above. As mentioned above, the first abnormal discharge detection model includes a first multi-head attention sub-model and a first deep convolutional neural network sub-model. The original biomedical signal to be detected can be input into the first multi-head attention sub-model in the first abnormal discharge detection model, and the first feature of the biomedical signal to be detected can be extracted through the first multi-head attention sub-model.

[0106] In an exemplary embodiment, the QKV of the first multi-head attention sub-model is the same, that is, they are all original signals of the biomedical signal to be detected. Among them, Q, K, and V correspond to the query (Query), key (Key), and value (Value) matrices in the multi-head attention model respectively.

[0107] The specific structure of the first multi-head attention sub-model in the present disclosure can refer to the structure of the multi-head attention model in the existing natural language processing field, and will not be repeated here.

[0108] After the original signal of the biomedical signal to be detected passes through the first multi-head attention sub-model, the first feature output by the first multi-head attention sub-model is a shallow feature. This shallow biomedical signal feature contains time information features, but it cannot represent the deeper signal meaning of the biomedical signal. Therefore, in the present disclosure, inspired by the field of computer vision, the shallow features are input into the deep convolutional neural network to extract deeper features of the biomedical signal, thereby improving the accuracy of abnormal discharge detection.

[0109] In step S520, the first feature of the biomedical signal to be detected is input into the first deep convolutional neural network sub-model in the first abnormal discharge detection model, and the abnormal discharge detection result of the object under test is obtained according to the output of the first deep convolutional neural network sub-model.

[0110] In an exemplary embodiment, as described above, the first deep convolutional neural network sub-model may include any one of a first deep dilated convolutional network sub-model and a first multi-layer convolutional neural network sub-model.

[0111] In the case where the first deep convolutional neural network model is the first deep expanded convolutional network sub-model, since the original biomedical signal, such as the above-mentioned 29 times 2000 EEG signal, is not very sensitive to convolution, the convolution expansion rate of each layer in the first deep expanded convolutional network sub-model can be the same, such as 1, that is, the first deep expanded convolutional network sub-model can be an ordinary multi-layer convolutional neural network sub-model, so that the processing efficiency of the model can be improved while ensuring the processing effect. Of course, the expansion rates of each layer of the first deep expanded convolutional network sub-model can also be different, and this exemplary embodiment does not make any special limitations on this.

[0112] The number of layers of the first deep dilated convolutional network sub-model and the dilation rate of each layer can be determined according to requirements or experimental results. For example, experimental results show that when the number of layers is 3 and the dilation rate of each layer is 1, the training result is best. In this case, the number of layers of the first deep dilated convolutional network sub-model can be 3, and the dilation rate of each layer can be 1.

[0113] In the case where the first deep convolutional neural network submodel is a first multi-level convolutional neural network submodel, the convolution kernels of each layer of the first multi-level convolutional neural network submodel may be the same, but the number of convolution kernels of each layer is different. The convolution kernels of each layer of the first multi-level convolutional neural network submodel may also be different, and this exemplary embodiment does not specifically limit this.

[0114] Exemplarily, an implementation of step S520 may include: inputting a first feature of the biomedical signal to be detected into a first deep convolutional neural network submodel in a first abnormal discharge detection model, and obtaining a second feature of the biomedical signal to be detected according to an output of the first deep dilated convolutional network submodel; inputting the second feature of the biomedical signal to be detected into a first fully connected layer, and obtaining a first fully connected feature according to an output of the first fully connected layer; processing the first fully connected feature by a preset activation function, and obtaining an abnormal discharge detection result of the object under test according to the processing result.

[0115] For example, the structure of the first abnormal discharge detection model can be as follows: Figure 6 As shown, the first abnormal discharge detection model is composed of a first multi-head attention sub-model 61, a first deep convolutional neural network sub-model 62 and a first fully connected layer 63.

[0116] In actual abnormal discharge detection, the first category of biomedical signals to be detected can be input into the first multi-head attention sub-model 61, and the biomedical signals to be detected are processed by the first multi-head attention sub-model. The shallow features of the biomedical signals to be detected are obtained according to the output of the first multi-head attention sub-model, and then the shallow features are input into the first deep convolutional neural network sub-model 62. The deeper features of the biomedical signals are extracted based on the first deep convolutional neural network sub-model, and the deep features are obtained according to the output of the first deep convolutional neural network sub-model. The deep features are then reduced in dimension using the first fully connected layer 63, and the features after dimensionality reduction of the fully connected layer are mapped by a preset activation function, such as a binary classification activation function sigmoid, to obtain an abnormal probability. When the abnormal probability is greater than a preset value, it is determined that the biomedical signal to be detected has abnormal discharge, otherwise it is determined that there is no abnormal discharge, thereby obtaining an abnormal discharge detection result of the biomedical signal to be detected. As mentioned above, according to the aforementioned Figure 4 The method shown obtains the abnormal discharge detection result of the detected object according to the abnormal discharge detection result of the biomedical signal to be detected.

[0117] In an exemplary embodiment, when the first abnormal discharge detection model determines that abnormal discharge exists, it is determined that abnormal discharge does exist in the biomedical signal to be detected, that is, the abnormal discharge detection result of the biomedical signal to be detected is abnormal. When the first discharge detection model determines that there is no abnormal discharge, it is determined that the biomedical signal to be detected is suspected abnormal discharge, and no suspected abnormal discharge label is added thereto. The doctor can manually confirm the abnormality of the biomedical signal to be detected with the suspected abnormal discharge label.

[0118] Exemplarily, when the biomedical signal to be detected is of the second category, Figure 7 The method shown in the figure obtains the abnormal discharge detection result of the object under test. Figure 7 The method for determining the abnormal discharge detection result of the second category of the biomedical signal to be detected may include steps S710 to S730. Among them:

[0119] In step S710, the biomedical signal to be detected is transformed in form to obtain image-like biomedical signal data corresponding to the biomedical signal to be detected.

[0120] In an exemplary embodiment, the biomedical signal to be detected includes a first matrix, and the first matrix corresponds to a first matrix dimension. For example, the biomedical signal to be detected can be represented by the first matrix, and the first matrix can be a matrix of N rows and M columns, such as the above-mentioned matrix of 29 rows and 2000 columns. N represents the number of channels of the biomedical signal, and 2000 represents the number of samples of each channel of the biomedical signal to be detected.

[0121] Based on this, exemplarily, an implementation of step S710 may include: performing matrix deformation on the first matrix to obtain a second matrix, the second matrix corresponds to a second matrix dimension, and the difference between the number of rows and the number of columns of the second matrix dimension is less than a preset value; and determining image-like biomedical signal data based on the second matrix.

[0122] For example, a first matrix can be converted into a second matrix with K rows and L columns through a matrix transformation function, such as a reshape function, wherein the absolute value of the difference between K and L is less than a preset value. The preset value can be customized according to needs, such as 0, 50, 100, etc. The preset value cannot be set too large, otherwise the difference in form between the converted matrix and the image data matrix will still be large. The purpose of setting the preset value is to make the converted matrix more suitable for convolution operations, such as converting the above-mentioned first matrix of 29 times 2000 into a second matrix of 290 times 200. In this way, it can be ensured that the form of the second matrix is ​​similar to that of the image signal, and the second matrix is ​​used as the image-like biomedical signal data of the biomedical signal to be detected.

[0123] It should be noted that in the process of converting from the first matrix to the second matrix, only the row and column dimensions of the matrix change, while the elements in the matrix do not change at all.

[0124] In step S720, the image-like biomedical signal data is input into the second deep convolutional neural network sub-model, and the abnormal discharge detection result of the subject is obtained according to the output of the second deep convolutional neural network sub-model.

[0125] In an exemplary embodiment, when the biomedical signal to be detected is of the second category, the target abnormal discharge detection model is the second abnormal discharge detection model described above. As described above, the second abnormal discharge detection model includes a second deep convolutional neural network sub-model. Image-like biomedical signal data, such as the second matrix described above, can be input into the second deep convolutional neural network sub-model, and deep features of the image-like biomedical signal data can be extracted by the second deep convolutional neural network sub-model, and whether the biomedical signal to be detected is really abnormal can be determined based on deeper features.

[0126] In an exemplary embodiment, as described above, the second deep convolutional neural network sub-model may include any one of a second deep dilated convolutional network sub-model and a second deep multi-level convolutional neural network sub-model.

[0127] In the case where the second deep convolutional neural network sub-model is a second deep dilated convolutional network sub-model, the second deep dilated convolutional network sub-model may be a multi-layer dilated convolutional neural network with different dilation rates for each layer, such as a 3-layer dilated convolutional network with dilation rates of 1, 2, and 3. In this way, through the second deep dilated convolutional network sub-model, deep feature extraction can be performed on image-like biomedical signal data to obtain richer biomedical signal features.

[0128] In the case where the second deep convolutional neural network sub-model is a second multi-level convolutional neural network sub-model, the convolution kernels of each layer of the second multi-level convolutional neural network sub-model may be different, such as the first layer includes three convolution kernels 1, the second layer includes two convolution kernels 2, and the third layer includes one convolution kernel 3, wherein convolution kernel 1, convolution kernel 2 and convolution kernel 3 are different, and this exemplary embodiment does not make any special limitation on this.

[0129] Exemplarily, a specific implementation of step S730 may include: inputting the image-like biomedical signal data into a second deep convolutional neural network sub-model, and obtaining a third feature of the image-like biomedical signal data based on an output of the second deep dilated convolutional network sub-model; inputting the third feature of the image-like biomedical signal data into a second fully connected layer, and obtaining a second fully connected feature based on an output of the second fully connected layer; processing the second fully connected feature by a preset activation function, and obtaining an abnormal discharge detection result of the subject based on the processing result.

[0130] Exemplarily, other implementations of step S720 may refer to the relevant contents of the above-mentioned step S520, which will not be described in detail here.

[0131] For example, when the second abnormal discharge detection model determines that the biomedical signal to be detected has abnormal discharge, then it is determined that the biomedical signal to be detected really has abnormal discharge. When the second abnormal discharge detection model determines that the biomedical signal to be detected does not have abnormal discharge, then it is determined that the biomedical signal to be detected is suspected of having abnormal discharge, and a suspected abnormal discharge label is added to the biomedical signal to be detected. The doctor can manually confirm the abnormality of the biomedical signal to be detected with the suspected abnormal discharge label, that is, the doctor determines whether the biomedical signal to be detected with the suspected abnormal label really has an abnormality. If the doctor believes that it is abnormal, then it is really abnormal; if the doctor believes that it is not abnormal, then it is not abnormal.

[0132] In another exemplary embodiment, for the biomedical signal to be detected that is determined to be the third category in step S210, that is, a normal biomedical signal to be detected, it can be directly determined that there is no abnormal discharge. Since the normal biomedical signal to be detected has low time dependence information, it can also be converted into image-like data and then input into the second abnormal discharge detection model to determine whether it is really normal through the second abnormal discharge detection model. If the second abnormal discharge detection model determines that it is normal, it is determined that it is really normal. Otherwise, it is determined to be a suspected normal biomedical signal to be detected, and a suspected normal label is added to it. The doctor can manually confirm the normality of the biomedical signal to be detected with the suspected normal label.

[0133] In the present disclosure, biomedical signals are classified according to their characteristics, and based on the classification results, suspected abnormal biomedical signals and the categories of suspected abnormalities can be preliminarily and quickly determined. For categories of biomedical signals that are more suitable for feature extraction using deep neural networks, they can be converted into image-like biomedical signal data and then processed using the corresponding abnormal discharge detection model, thereby fully tapping the advantages of deep neural network models in biomedical signal processing, more accurately confirming abnormal situations, and making the detection results of abnormal discharges of biomedical signals more accurate and reliable, thereby providing doctors with more accurate auxiliary diagnostic information.

[0134] In addition, it should be noted that the above-mentioned figures are only schematic illustrations of the processes included in the method according to an exemplary embodiment of the present invention, and are not intended to be limiting. It is easy to understand that the processes shown in the above-mentioned figures do not indicate or limit the time sequence of these processes. In addition, it is also easy to understand that these processes can be performed synchronously or asynchronously, for example, in multiple modules.

[0135] Exemplary Devices

[0136] The exemplary embodiment of the present disclosure also provides an abnormal discharge detection device. Figure 8 As shown, the abnormal discharge detection device 800 may include the following program modules: a target category determination module 810, configured to obtain the biomedical signal to be detected of the subject, and determine the target category of the biomedical signal to be detected; a model selection module 820, configured to determine the target abnormal discharge detection model corresponding to the target category from multiple pre-trained abnormal discharge detection models according to the target category; an abnormality detection module 830, configured to process the biomedical signal to be detected based on the target abnormal discharge detection model, and obtain the abnormal discharge detection result of the subject according to the processing result.

[0137] In an exemplary embodiment, obtaining the biomedical signals to be detected of the subject includes: obtaining the collected biomedical signals of the subject; obtaining the biomedical signals of the subject in segments according to preset time lengths, and determining at least one biomedical signal to be detected of the subject.

[0138] In an exemplary embodiment, the abnormality detection module 830 can be specifically configured as follows: based on the target abnormal discharge detection model, processing the biomedical signal to be detected, and obtaining the abnormal discharge detection result of the biomedical signal to be detected according to the processing result; when the abnormal discharge detection result of the biomedical signal to be detected is abnormal discharge, marking the biomedical signal to be detected; and obtaining the abnormal discharge detection result of the subject according to the marked biomedical signal to be detected.

[0139] In an exemplary embodiment, when the target category is the first category, the target abnormal discharge detection model is a first abnormal discharge detection model; based on the target abnormal discharge detection model, the biomedical signal to be detected is processed, and the abnormal discharge detection result of the object under test is obtained according to the processing result, including: inputting the biomedical signal to be detected into a first multi-head attention sub-model in the first abnormal discharge detection model, and obtaining a first feature of the biomedical signal to be detected according to the output of the first multi-head attention sub-model; inputting the first feature into a first deep convolutional neural network sub-model in the first abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the output of the first deep convolutional neural network sub-model.

[0140] In an exemplary embodiment, when the target category is the second category, the target abnormal discharge detection model is a second abnormal discharge detection model, and the second abnormal discharge detection model includes a second deep convolutional neural network sub-model; based on the target abnormal discharge detection model, the biomedical signal to be detected is processed, and the abnormal discharge detection result of the object under test is obtained according to the processing result, including: performing a form transformation on the biomedical signal to be detected to obtain image-like biomedical signal data corresponding to the biomedical signal to be detected; inputting the image-like biomedical signal data into the second deep convolutional neural network sub-model, and obtaining the abnormal discharge detection result of the object under test according to the output of the second deep convolutional neural network sub-model.

[0141] In an exemplary embodiment, the biomedical signal to be detected includes a first matrix, which corresponds to a first matrix dimension. The biomedical signal to be detected of the object under test is transformed in form to obtain image-like biomedical signal data corresponding to the biomedical signal to be detected; the first matrix is ​​matrix deformed to obtain a second matrix, which corresponds to a second matrix dimension, and the difference between the number of rows and the number of columns of the second matrix dimension is less than a preset value; and the image-like biomedical signal data is determined based on the second matrix.

[0142] In an exemplary embodiment, any of the deep convolutional neural network sub-models includes any of a deep dilated convolutional network sub-model and a multi-layer convolutional neural network sub-model.

[0143] The specific details of each part of the above-mentioned device have been described in detail in the implementation of the corresponding method part above. The undisclosed details can be found in the implementation of the method part above, and will not be repeated here.

[0144] Exemplary Storage Media

[0145] The storage medium according to the exemplary embodiment of the present invention is described below.

[0146] In this exemplary embodiment, the above method can be implemented by a program product, such as a portable compact disk read-only memory (CD-ROM) and including program code, and can be run on a device, such as a personal computer. However, the program product of the present invention is not limited thereto. In this document, a readable storage medium can be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device.

[0147] The program product may adopt any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof.

[0148] Computer readable signal media may include data signals propagated in baseband or as part of a carrier wave, in which readable program code is carried. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Readable signal media may also be any readable medium other than a readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0149] The program code contained on the readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wired, optical cable, RE, etc., or any suitable combination of the foregoing.

[0150] Program code for performing the operations of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user computing device, 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., using an Internet service provider to connect through the Internet).

[0151] Exemplary computer program products

[0152] The exemplary embodiments of the present disclosure also provide a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the abnormal discharge detection method described above is implemented.

[0153] In one embodiment, the computer program product may be a tangible product containing a computer program, such as a computer-readable storage medium storing a computer program. The readable storage medium may be a storage medium based on electrical, magnetic, optical, electromagnetic, infrared, or 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 implemented as a non-volatile storage medium storing a computer program, such as a read-only memory, a NAND flash memory (Nand Flash), and the like.

[0154] In one embodiment, the computer program product may be an intangible product including a computer program. Exemplarily, the computer program product may be implemented as a virtual digital product, such as a digital file storing an executable file, an installation package, etc. of the computer program.

[0155] The code of the computer program can be written in one or more programming languages. Programming languages ​​such as C language, Java, C++, Python, etc. The program code can be executed completely 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 the remote computing device, or completely on the 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 (for example, through an Internet connection provided by an operator).

[0156] 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, and then run computer programs. When a computer program runs on an electronic device, its code is used to enable the electronic device to execute (more specifically, it can enable the processor of the electronic device to execute) the method steps of various exemplary embodiments of the present disclosure, such as the abnormal discharge detection method described above, which includes the following steps: obtaining a biomedical signal to be detected of the subject, and determining a target category of the biomedical signal to be detected; according to the target category, determining a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models; based on the target abnormal discharge detection model, processing the biomedical signal to be detected, and obtaining an abnormal discharge detection result of the subject according to the processing result.

[0157] By executing the above method steps through a computer program, on the one hand, according to the abnormal discharge detection model, the automatic detection and identification of abnormal discharge of biomedical signals can be realized, thereby improving the efficiency of abnormal discharge detection; on the other hand, according to the category of the biomedical signal, the biomedical signal with suspected abnormality is determined, and the target abnormal discharge detection model corresponding to the category of the suspected abnormal biomedical signal is selected, and the biomedical signals of different categories are processed in a targeted manner to improve the accuracy of abnormal discharge detection of biomedical signals; on the other hand, more accurate reference information can be provided for doctors, so that doctors do not need to observe all the data in detail, but only need to roughly browse some key information, and then combine it with the abnormal discharge detection results to quickly and accurately determine the diagnosis results of the patient, saving the workload of doctors and facilitating wide application in clinical practice.

[0158] Exemplary Electronic Devices

[0159] refer to Fig. 9An electronic device of an exemplary embodiment of the present disclosure is described. The electronic device is the terminal device 110 or the server 120 mentioned above. The electronic device may include a processor and a memory. The memory stores executable instructions of the processor, such as a computer program. The processor executes the method steps of various exemplary embodiments of the present disclosure by executing the executable instructions. In addition, the electronic device may also include a display for displaying a graphical user interface.

[0160] Reference below Fig. 9 , the electronic device is exemplarily described in the form of a general-purpose computing device. It should be understood that Fig. 9 The electronic device 900 shown is only an example and should not limit the functions and scope of use of the embodiments of the present disclosure.

[0161] like Fig. 9 As shown, the electronic device 900 may include: a processor 910 , a memory 920 , a bus 930 , an I / O (input / output) interface 940 , a network adapter 950 , and a display 960 .

[0162] The memory 920 may include a volatile memory, such as a RAM 921, a cache unit 922, and may also include a non-volatile memory, such as a 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 combination thereof may include the implementation of a network environment. For example, the program module 924 may include each module in the above-mentioned device.

[0163] The processor 910 may include one or more processing units. For example, the processor 910 may include 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), etc.

[0164] The processor 910 can be used to execute executable instructions stored in the memory 920, such as the above-mentioned abnormal discharge detection method, which includes the following steps: obtaining a biomedical signal to be detected from the subject, and determining a target category of the biomedical signal to be detected; according to the target category, determining a target abnormal discharge detection model corresponding to the target category from multiple pre-trained abnormal discharge detection models; based on the target abnormal discharge detection model, processing the biomedical signal to be detected, and obtaining an abnormal discharge detection result of the subject according to the processing result.

[0165] The above method is implemented through a computer program. On the one hand, according to the abnormal discharge detection model, the automatic detection and identification of abnormal discharges of biomedical signals can be realized, thereby improving the efficiency of abnormal discharge detection; on the other hand, according to the category of the biomedical signal, a target abnormal discharge detection model corresponding to the category of the biomedical signal is selected, and targeted processing is performed on biomedical signals of different categories to improve the accuracy of abnormal discharge detection of biomedical signals; on the other hand, more accurate reference information can be provided for doctors, so that doctors do not need to observe all the data in detail, but only need to roughly browse some key information, and then combine it with the abnormal discharge detection results to quickly and accurately determine the diagnosis results of the patient, saving the workload of doctors and facilitating wide application in clinical practice.

[0166] The bus 930 is used to realize the connection between different components of the electronic device 900, and may include a data bus, an address bus, and a control bus.

[0167] The electronic device 900 can communicate with one or more external devices 1000 (eg, a keyboard, a mouse, an external controller, etc.) through the I / O interface 940 .

[0168] The electronic device 900 can 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.

[0169] The electronic device 900 may display a graphical user interface, such as a graphical user interface displaying abnormal discharge detection results, through the display 960 .

[0170] 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: microcode, device drivers, redundant processors, external disk drive arrays, RAID systems, tape drives, and data backup storage systems, etc.

[0171] It should be understood that the present disclosure is not limited to the specific method steps or structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from the scope thereof. Those skilled in the art will easily think of other embodiments based on the specific embodiments provided by the present disclosure. Therefore, the specific embodiments provided by the present disclosure are only exemplary, and the scope and spirit of the present disclosure are indicated by the claims, and any variations, uses or adaptive changes of the present disclosure should be covered, which follow the general principles of the present disclosure and include common knowledge or customary technical means in the technical field that are not disclosed in the present disclosure.

Claims

1. An abnormal discharge detection method, characterized in that: include: Acquire a biomedical signal to be detected from a subject, and determine a target category of the biomedical signal to be detected; According to the target category, determining a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models; Based on the target abnormal discharge detection model, the biomedical signal to be detected is processed, and an abnormal discharge detection result of the detected object is obtained according to the processing result; The target categories include a first category and a second category, the first category characterizing suspected abnormal biomedical signals whose to-be-detected biomedical signals are dependent on time information, and the second category includes suspected abnormal biomedical signals whose to-be-detected biomedical signals are not dependent on time information; Among them, the processing of the biomedical signal to be detected based on the target abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the processing result includes: when the target category is the first category, directly inputting the biomedical signal to be detected into the first abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the output of the first abnormal discharge detection model, and the first abnormal discharge detection model includes a first multi-head attention sub-model and a first deep convolutional neural network sub-model; when the target category is the second category, transforming the form of the biomedical signal to be detected to obtain image-like biomedical signal data corresponding to the biomedical signal to be detected, inputting the image-like biomedical signal data into the second abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the output of the second abnormal discharge detection model, and the second abnormal discharge detection model includes a second deep convolutional neural network sub-model.

2. The method according to claim 1, characterized in that The step of obtaining the biomedical signal to be detected from the subject comprises: Acquiring the collected biomedical signals of the subject; The biomedical signals of the subject are acquired in segments according to a preset time length, and at least one biomedical signal to be detected of the subject is determined.

3. The method according to claim 2, characterized in that The processing of the biomedical signal to be detected based on the target abnormal discharge detection model and obtaining the abnormal discharge detection result of the detected object according to the processing result comprises: Based on the target abnormal discharge detection model, the biomedical signal to be detected is processed, and an abnormal discharge detection result of the biomedical signal to be detected is obtained according to the processing result; When the abnormal discharge detection result of the biomedical signal to be detected is abnormal discharge, marking the biomedical signal to be detected; The abnormal discharge detection result of the detected object is obtained according to the marked biomedical signal to be detected.

4. The method according to claim 1, characterized in that The step of directly inputting the biomedical signal to be detected into a first abnormal discharge detection model and obtaining an abnormal discharge detection result of the detected object according to an output of the first abnormal discharge detection model comprises: Inputting the biomedical signal to be detected into a first multi-head attention sub-model in the first abnormal discharge detection model, and obtaining a first feature of the biomedical signal to be detected according to an output of the first multi-head attention sub-model; The first feature is input into a first deep convolutional neural network sub-model in the first abnormal discharge detection model, and an abnormal discharge detection result of the object under test is obtained according to an output of the first deep convolutional neural network sub-model.

5. The method according to claim 4, characterized in that The biomedical signal to be detected includes a first matrix, the first matrix corresponds to a first matrix dimension, and the form transformation of the biomedical signal to be detected to obtain the image-like biomedical signal data corresponding to the biomedical signal to be detected includes: Performing matrix deformation on the first matrix to obtain a second matrix, where the second matrix corresponds to a second matrix dimension, and a difference between the number of rows and the number of columns of the second matrix dimension is less than a preset value; The image-like biomedical signal data is determined according to the second matrix.

6. The method according to claim 1, characterized in that Any of the deep convolutional neural network sub-models includes any one of a deep dilated convolutional network sub-model and a multi-layer convolutional neural network sub-model.

7. An abnormal discharge detection device, characterized in that: include; A target category determination module is configured to obtain a biomedical signal to be detected from a subject and determine a target category of the biomedical signal to be detected; A model selection module is configured to determine, according to the target category, a target abnormal discharge detection model corresponding to the target category from a plurality of pre-trained abnormal discharge detection models; an abnormality detection module, configured to process the biomedical signal to be detected based on the target abnormal discharge detection model, and obtain an abnormal discharge detection result of the detected object according to the processing result; The target categories include a first category and a second category, the first category characterizing suspected abnormal biomedical signals whose to-be-detected biomedical signals are dependent on time information, and the second category includes suspected abnormal biomedical signals whose to-be-detected biomedical signals are not dependent on time information; Among them, the processing of the biomedical signal to be detected based on the target abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the processing result includes: when the target category is the first category, directly inputting the biomedical signal to be detected into the first abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the output of the first abnormal discharge detection model, and the first abnormal discharge detection model includes a first multi-head attention sub-model and a first deep convolutional neural network sub-model; when the target category is the second category, transforming the form of the biomedical signal to be detected to obtain image-like biomedical signal data corresponding to the biomedical signal to be detected, inputting the image-like biomedical signal data into the second abnormal discharge detection model, and obtaining the abnormal discharge detection result of the object under test according to the output of the second abnormal discharge detection model, and the second abnormal discharge detection model includes a second deep convolutional neural network sub-model.

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: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the one or more processors to implement the method according to any one of claims 1 to 6.

Images