Medical electroencephalogram data set construction method based on negative example optimization
By using positive example category center distance as the guide to select negative example samples in the EEG data set construction, the problems of low efficiency and poor quality of medical EEG data set construction in the prior art are solved, and efficient and automated construction of medical EEG data sets are achieved. Methods of datasets.
Patent Information
- Application Number
- CN202411547119.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-01
- Publication Date
- 2025-01-10
AI Technical Summary
The lack of automation support in the prior art has led to low efficiency in the construction of medical EEG data sets and inability to guarantee quality, especially the lack of effective methods when selecting negative example samples.
When extracting positive EEG samples, the negative example samples are preferred based on the distance of the positive example category center as the guide, and the positive example category center and distance standards are updated, so as to automatically construct the medical EEG data set while ensuring the quality of negative examples. The specific steps include obtaining user parameters, sorting positive example label sets, intercepting EEG signal tensors and adding them to the sample set, randomly generating negative example samples and updating the distance standard until the construction is completed.
It realizes efficient and automated construction of medical EEG data sets while ensuring the quality of negative cases, fills the gaps in the existing technology, and improves the efficiency and quality of data set construction.
Smart Images
Figure CN119293508A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical fields of electroencephalogram (EEG) data, medical big data, etc., and specifically relates to a method for constructing a medical EEG data set based on negative example optimization. Background Art
[0002] Medical EEG data contains information about various diseases and is used for the diagnosis of brain diseases. It is also a resource for studying the mechanisms of brain diseases. To conduct research using medical EEG data, it is necessary to first manually label the EEG signals, and then extract EEG samples from the EEG signals based on the labels to construct an EEG dataset. Constructing an EEG dataset requires both manually labeled positive examples and manually unlabeled negative examples. What kind of EEG signal segments to select as negative examples is a problem that must be faced in the automated construction of medical EEG datasets. The existing technology does not provide automated support for constructing medical EEG datasets. The current situation is that the efficiency of constructing medical EEG datasets is very low, and the quality of constructed medical EEG datasets is not guaranteed. Summary of the invention
[0003] In view of the gaps and deficiencies in the prior art and the actual needs for constructing a medical EEG data set, the present invention provides a method for constructing a medical EEG data set based on negative example optimization.
[0004] Each time this method extracts a positive sample from the EEG signal, within the time range between the last positive sample and the current positive sample, the distance from the center of each positive class is used as a guide to select the negative sample, and the center and distance standard of each positive class are updated, so that the EEG signal segments with large differences from the positive samples can be selected as negative samples and added to the EEG data set while extracting EEG positive samples based on manual annotation. This method can realize the automatic construction of medical EEG data sets while ensuring the quality of negative samples.
[0005] The technical solution specifically adopted by the present invention to solve the technical problem is: Each time a positive sample is extracted from the EEG signal, within the time range between the previous positive sample and the current positive sample, the negative sample is selected based on the distance from the center of each positive category, and the center and distance standard of each positive category are updated. This allows the EEG positive samples to be extracted based on the annotations while the EEG signal segments with large differences from the positive samples are selected as negative samples and added to the EEG data set.
[0006] Furthermore, the method specifically includes the following steps: Step 1: Obtain the parameter N (N:1 represents the ratio of negative examples to positive examples, N is a positive integer) provided by the user, the EEG file and the positive example annotation set corresponding to the EEG file, and sort the positive example annotation set in chronological order; Step 2: Take out the first positive example annotation in the positive example annotation set, extract the EEG signal from the EEG file according to the start and end time of the annotation to obtain a tensor, add the tensor to the sample set of the corresponding category according to the labeled category, and set the tensor as the center of the category; set the distance standard D to 0; set the time range Range to the start time of the EEG file to the start time of the annotation; Step 3: Based on D, Range and the center of each class, extract N negative samples from the EEG file and add these N negative samples to the negative sample set; Step 4: If the positive example annotation set does not have the next positive example annotation, go to step 5; otherwise, take the next positive example annotation, extract the EEG signal from the EEG file according to the start and end time of the current annotation to obtain a tensor, add the tensor to the sample set of the category according to the current annotation category, update the center of the category with the average of all samples of the category, update the start time of the time range Range with the end time of the previous sample, update the end time of the time range Range with the start time of the current sample, and return to step 3; Step 5: Output the medical EEG dataset obtained from the EEG file.
[0007] Furthermore, step 3 specifically includes the following steps: Step 3.1: Set n=1; Step 3.2: If n>N, output N negative samples and end; Step 3.3: Randomly generate a negative example annotation within the time range specified by Range, intercept a tensor in the EEG signal according to the start and end time of the negative example annotation, calculate the distance between the tensor and the center of each positive example category, and take the smallest one as d min ; Step 3.4: If d min ≤D , go to step 3.3; otherwise, go to step 3.5; Step 3.5: Confirm that the tensor is a new negative sample and use d min Update D, update n with n+1, and go to step 3.2.
[0008] Furthermore, the tensor is a first-order tensor or a second-order tensor: when the intercepted EEG signal is a signal of one lead, the tensor is first-order; when the intercepted EEG signal is a signal of multiple leads, the tensor is second-order; during the execution of the medical EEG dataset construction method once, the tensor is constant to the first-order or second-order.
[0009] Furthermore, the distance is one of Euclidean distance, Manhattan distance or Chebyshev distance.
[0010] Compared with the prior art, the present invention and its preferred solution can realize the efficient and automated construction of medical EEG data sets while ensuring the quality of negative examples, thus filling the gap in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments: Figure 1 The figure is a flow chart of a method according to an embodiment of the present invention. DETAILED DESCRIPTION
[0012] In order to make the features and advantages of the present invention more clearly understood, the following embodiments are specifically described in detail as follows: It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in this specification have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.
[0013] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.
[0014] like Figure 1 As shown, the specific design of the method for constructing a medical EEG dataset based on negative example optimization provided by the present invention includes the following steps: Step 1: Obtain the parameter N (N:1 represents the ratio of negative examples to positive examples, N is a positive integer) provided by the user, the EEG file and the positive example annotation set corresponding to the EEG file, and sort the positive example annotation set in chronological order; Step 2: Take out the first positive example annotation in the positive example annotation set, extract the EEG signal from the EEG file according to the start and end time of the annotation to obtain a tensor, add the tensor to the sample set of the category according to the annotation category, and set the tensor as the center of the category; set the distance standard D to 0; set the time range Range to the start time of the EEG file to the start time of the annotation; Step 3: Based on D, Range and the center of each class, extract N negative samples from the EEG file and add these N negative samples to the negative sample set; Step 4: If the positive example annotation set does not have the next positive example annotation, go to step 5; otherwise, take the next positive example annotation, extract the EEG signal from the EEG file according to the start and end time of the current annotation to obtain a tensor, add the tensor to the sample set of the category according to the current annotation category, update the center of the category with the average of all samples of the category, update the start time of the time range Range with the end time of the previous sample, update the end time of the time range Range with the start time of the current sample, and return to step 3; Step 5: Output the medical EEG dataset obtained from the EEG file specified by the user.
[0015] As a preferred solution of this embodiment, the specific implementation process of step 3 can be described as follows: Step 3.1: Set n=1; Step 3.2: If n>N, output N negative samples and end; Step 3.3: Randomly generate a negative example annotation within the time range specified by Range, intercept a tensor in the EEG signal according to the start and end time of the negative example annotation, calculate the distance between the tensor and the center of each positive example category, and take the smallest one as d min ; Step 3.4: If d min ≤D , go to step 3.3; otherwise, go to step 3.5; Step 3.5: Confirm that the tensor is a new negative sample and use d min Update D, update n with n+1, and go to step 3.2.
[0016] The method constructed above can be used to extract positive and negative samples from an EEG file with positive examples to construct an EEG dataset. When there are multiple labeled EEG files, this method can be repeatedly used to extract samples from each EEG file to construct an EEG dataset.
[0017] The tensor described above is a segment of EEG signal intercepted from the EEG file according to the time range specified by the annotation. The tensor here can be first-order or second-order. When the intercepted EEG signal is the signal of one lead, the tensor is first-order. When the intercepted EEG signal is the signal of multiple leads, the tensor is second-order. However, it should be noted that in one application of this method, the tensor is either first-order or second-order, and cannot be first-order sometimes and second-order sometimes.
[0018] The distance described in this solution can be various distances such as Euclidean distance, Manhattan distance, Chebyshev distance, etc., but only one distance can be selected in one application. As a preferred embodiment, Euclidean distance can be used.
[0019] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.
[0020] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0021] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0022] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0023] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0024] The above is only a preferred embodiment of the present invention, and does not limit the present invention in other forms. Any technician familiar with the profession may use the above disclosed technical content to change or modify it into an equivalent embodiment with equivalent changes. However, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the technical solution of the present invention still belongs to the protection scope of the technical solution of the present invention.
[0025] The present invention is not limited to the above-mentioned optimal implementation mode. Anyone can derive other various forms of medical EEG data set construction methods based on negative example optimization under the inspiration of the present invention. All equivalent changes and modifications made according to the scope of the patent application of the present invention should fall within the scope of the present invention.
Claims
1. A method for constructing a medical EEG dataset based on negative example optimization, characterized in that: Each time a positive sample is extracted from the EEG signal, within the time range between the previous positive sample and the current positive sample, the negative sample is selected based on the distance from the center of each positive category, and the center and distance standard of each positive category are updated. This allows the EEG positive samples to be extracted based on the annotations while the EEG signal segments with large differences from the positive samples are selected as negative samples and added to the EEG data set.
2. The method for constructing a medical EEG dataset based on negative example optimization according to claim 1, characterized in that: The specific steps include: Step 1: Obtain the parameter N (N:1 represents the ratio of negative examples to positive examples, N is a positive integer) provided by the user, the EEG file and the positive example annotation set corresponding to the EEG file, and sort the positive example annotation set in chronological order; Step 2: Take out the first positive example annotation in the positive example annotation set, extract the EEG signal from the EEG file according to the start and end time of the annotation to obtain a tensor, add the tensor to the sample set of the corresponding category according to the labeled category, and set the tensor as the center of the category; set the distance standard D to 0; set the time range Range to the start time of the EEG file to the start time of the annotation; Step 3: Based on D, Range and the center of each class, extract N negative samples from the EEG file and add these N negative samples to the negative sample set; Step 4: If the positive example annotation set does not have the next positive example annotation, go to step 5; otherwise, take the next positive example annotation, extract the EEG signal from the EEG file according to the start and end time of the current annotation to obtain a tensor, add the tensor to the sample set of the category according to the current annotation category, update the center of the category with the average of all samples of the category, update the start time of the time range Range with the end time of the previous sample, update the end time of the time range Range with the start time of the current sample, and return to step 3; Step 5: Output the medical EEG dataset obtained from the EEG file.
3. The method for constructing a medical EEG dataset based on negative example optimization according to claim 2, characterized in that: Step 3 specifically includes: Step 3.1: Set n=1; Step 3.2: If n>N, output N negative samples and end; Step 3.3: Randomly generate a negative example annotation within the time range specified by Range, intercept a tensor in the EEG signal according to the start and end time of the negative example annotation, calculate the distance between the tensor and the center of each positive example category, and take the smallest one as d min ; Step 3.4: If d min ≤D , go to step 3.3; otherwise, go to step 3.5; Step 3.5: Confirm that the tensor is a new negative sample and use d min Update D, update n with n+1, and go to step 3.
2.
4. The method for constructing a medical EEG dataset based on negative example optimization according to claim 2, characterized in that: The tensor is a first-order tensor or a second-order tensor: when the intercepted EEG signal is a signal of one lead, the tensor is first-order; when the intercepted EEG signal is a signal of multiple leads, the tensor is second-order; during the execution of the medical EEG data set construction method once, the tensor is constant to the first order or the second order.
5. The method for constructing a medical EEG dataset based on negative example optimization according to claim 1, characterized in that: The distance is one of Euclidean distance, Manhattan distance or Chebyshev distance.