A wireless intelligent angiography device imaging push intelligent control method and system

Abstract

The embodiment of the present invention provides a method and system for intelligent control of contrast injection of a wireless intelligent angiography device, which belongs to the field of artificial intelligence technology. The method includes: obtaining first operation data and a first contrast state; performing data recognition on the first operation data to obtain first abnormal data, and processing the first operation data according to the first abnormal data to obtain second operation data; obtaining a second contrast state from the first contrast state according to the second operation data; performing feature selection on the second operation data using neighborhood information to obtain target-related features; obtaining third operation data from the second operation data according to the target-related features; determining a state prediction model according to the third operation data and the second contrast state; obtaining initial operation data, and predicting the contrast injection state of the initial operation data according to the state prediction model to obtain a target prediction result; adjusting the initial operation data according to the target prediction result to obtain the target operation data of the target user.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent control method and system for radiographic injection of a wireless intelligent radiographic device. Background Art

[0002] Contrast agents play a vital role in medical imaging. They can significantly improve the clarity and contrast of images, thereby helping doctors diagnose diseases more accurately. However, the injection process of contrast agents must not only ensure image quality, but also minimize patient discomfort. Uniform injection of contrast agents can effectively reduce pain and discomfort in patients while improving image clarity, which is crucial to improving patients' overall experience and medical outcomes. Despite the many advantages of uniform injection of contrast agents, existing technologies still face major challenges in practical applications. Current technologies cannot make intelligent injection adjustments based on individual differences of different patients, which makes it impossible to guarantee the comfort of each patient when receiving contrast agent injection. For example, for patients with low weight or sensitivity to contrast agents, too fast an injection speed may cause nausea, vomiting or other uncomfortable symptoms; while for patients with heavy weight or good tolerance, too slow an injection speed may result in poor image quality. Existing technologies lack real-time monitoring and feedback mechanisms, and cannot adjust the injection plan according to the patient's real-time response. This means that during the injection process, doctors cannot promptly detect and correct problems that may cause patient discomfort, thereby increasing patients' pain and medical risks. Summary of the invention

[0003] The main purpose of the embodiments of the present invention is to provide a method and system for intelligent control of contrast injection of a wireless intelligent angiography device, aiming to solve the problem that the related technology lacks a real-time monitoring and feedback mechanism, and cannot adjust the injection plan according to the patient's real-time response, thereby increasing the patient's pain and medical risks.

[0004] In a first aspect, an embodiment of the present invention provides an intelligent control method for radiographic injection of a wireless intelligent radiographic device, comprising:

[0005] Obtaining first operation data corresponding to a historical user during the injection process and a first imaging state corresponding to the first operation data;

[0006] Performing abnormal data identification on the first operation data to obtain first abnormal data, and processing the first operation data according to the first abnormal data to obtain second operation data;

[0007] Obtaining a second imaging state corresponding to the second operating data from the first imaging state according to the second operating data;

[0008] Using the neighborhood information to perform feature selection on the second operation data to obtain target-related features corresponding to the injection process;

[0009] Obtaining third operation data corresponding to the target association feature from the second operation data according to the target association feature;

[0010] Determine a corresponding contrast injection state prediction model during the injection process according to the third operation data and the second contrast state;

[0011] Obtaining initial operation data corresponding to the target user during the injection process, and predicting the contrast bolus injection state of the initial operation data according to the contrast bolus injection state prediction model to obtain a target prediction result;

[0012] The initial operation data is adjusted according to the target prediction result to obtain the target operation data corresponding to the target user during the injection process.

[0013] In a second aspect, an embodiment of the present invention provides an intelligent control system for radiographic injection of a wireless intelligent radiographic device, comprising:

[0014] A data acquisition module, used for obtaining first operation data corresponding to the historical user during the injection process and a first contrast state corresponding to the first operation data;

[0015] a data processing module, configured to perform abnormal data identification on the first operation data to obtain first abnormal data, and process the first operation data according to the first abnormal data to obtain second operation data;

[0016] a first screening module, configured to obtain a second imaging state corresponding to the second operation data from the first imaging state according to the second operation data;

[0017] A feature analysis module, used to perform feature selection on the second operation data using neighborhood information to obtain target-related features corresponding to the injection process;

[0018] A second screening module, configured to obtain third operation data corresponding to the target association feature from the second operation data according to the target association feature;

[0019] A model determination module, in which a user determines a corresponding contrast injection state prediction model during the injection process according to the third operation data and the second contrast state;

[0020] A data prediction module, used for obtaining the initial operation data corresponding to the target user during the injection process, and predicting the contrast bolus injection state of the initial operation data according to the contrast bolus injection state prediction model to obtain a target prediction result;

[0021] The data adjustment module is used to adjust the initial operation data according to the target prediction result to obtain the target operation data corresponding to the target user during the injection process.

[0022] The embodiment of the present invention provides a method and system for intelligent control of contrast injection of a wireless intelligent contrast instrument, the method comprising: obtaining first operation data corresponding to the historical user during the injection process and the first contrast state corresponding to the first operation data, and then performing abnormal data identification on the first operation data to obtain the first abnormal data, and processing the first operation data according to the first abnormal data to obtain the second operation data; thereby, by performing abnormal data identification and processing on the first operation data, more accurate and reliable second operation data can be obtained, thereby improving the accuracy of subsequent analysis and prediction. Then, according to the second operation data, the second contrast state corresponding to the second operation data is obtained from the first contrast state; and the neighborhood information is used to perform feature selection on the second operation data to obtain the target-related features corresponding to the injection process; thereby, the neighborhood information is used to perform feature selection to obtain the target-related features most relevant to the injection process, which helps to simplify the model and improve the accuracy of the prediction. At the same time, this can also help understand which operating factors have the greatest impact on the contrast bolus state, and then obtain the third operating data corresponding to the target-related feature from the second operating data according to the target-related feature; thereby determining the corresponding contrast bolus state prediction model during the injection process according to the third operating data and the second contrast state; then obtain the corresponding initial operating data of the target user during the injection process, and predict the contrast bolus state of the initial operating data according to the contrast bolus state prediction model to obtain the target prediction result; finally, adjust the initial operating data according to the target prediction result to obtain the target operating data corresponding to the target user during the injection process. Thus, by predicting and adjusting the initial operating data of the target user, personalized target operating data can be provided for each target user, thereby improving the efficiency and safety of the injection process, thereby ensuring the comfort of the target user while ensuring the image quality, and solving the problem of the lack of real-time monitoring and feedback mechanism in the related technology, and the inability to adjust the bolus plan according to the patient's real-time response, thereby increasing the patient's pain and medical risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 A schematic flow chart of an intelligent control method for radiographic injection of a wireless intelligent radiographic device provided by an embodiment of the present invention;

[0025] Figure 2 A schematic diagram of the module structure of an intelligent control system for radiographic injection of a wireless intelligent radiographic device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may also be decomposed, combined or partially merged, so the actual execution order may change according to actual conditions.

[0028] It should be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0029] The embodiment of the present invention provides a method and system for intelligent control of contrast injection of a wireless intelligent angiography device. The method can be applied to a terminal device, which can be an electronic device such as a tablet computer, a laptop computer, a desktop computer, a personal digital assistant, and a wearable device. The terminal device can be a server or a server cluster.

[0030] Some embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0031] Please refer to Figure 1 , Figure 1 A flowchart of an intelligent control method for radiographic contrast injection of a wireless intelligent angiography device provided in an embodiment of the present invention.

[0032] like Figure 1 As shown, the wireless intelligent angiography instrument contrast injection intelligent control method includes steps S101 to S108.

[0033] Step S101, obtaining first operation data corresponding to a historical user during an injection process and a first angiography state corresponding to the first operation data.

[0034] Exemplarily, the first operation data corresponding to the historical user during the injection process is obtained from the database, and the first imaging state corresponding to the historical user under the first operation data is obtained, and the first imaging state includes but is not limited to clear imaging but the user is uncomfortable, clear imaging but the user is comfortable, unclear imaging and the user is uncomfortable, and unclear imaging but the user is comfortable. The first operation data includes but is not limited to pressure data, injection volume data, injection speed data, remaining liquid volume data, and pressure-speed curve of the historical user during the injection process.

[0035] Step S102: performing abnormal data identification on the first operation data to obtain first abnormal data, and processing the first operation data according to the first abnormal data to obtain second operation data.

[0036] Exemplarily, anomaly detection methods such as Isolation Forest and Local Outlier Factor (LOF) are used to identify abnormal data under each parameter in the first operation data to obtain the first abnormal data corresponding to each parameter, and then data fusion is performed on the first abnormal data corresponding to each parameter to obtain all abnormal positions corresponding to the first operation data, and then the data of the first operation data at the abnormal position is eliminated to obtain the second operation data.

[0037] In some embodiments, the performing abnormal data identification on the first operation data to obtain first abnormal data includes: determining a target window, and performing data division on the first operation data according to the target window to obtain first window data; performing data reconstruction on the first window data using a data reconstruction model to obtain first error information corresponding to the first window data; determining a first abnormal score corresponding to the first window data according to the first error information; and performing abnormal data identification on the first operation data according to the first abnormal score to obtain the first abnormal data.

[0038] Exemplarily, the target window is determined. The size and range of the target window can be set according to the characteristics and requirements of the first operation data, and can usually be determined by analyzing the time series characteristics of the data or business requirements.

[0039] Exemplarily, the first operation data is divided according to the determined target window according to the time period to obtain the first window data, wherein the first window data includes the target data corresponding to each parameter in the first operation data under the target window.

[0040] For example, the first operation data includes first corresponding data corresponding to the first parameter and second corresponding data corresponding to the second parameter. When the target window slides in the first corresponding data, it also slides in the second corresponding data at the same time, so that the data corresponding to the target window under the first corresponding data and the second corresponding data at the same time is determined as the first window data.

[0041] Exemplarily, normal operation data is obtained through data labeling, and then the normal operation data is compressed and reconstructed through a neural network model, and normal reconstructed data corresponding to the normal operation data is output. When the error between the normal operation data and the normal reconstructed data meets a preset condition, a data reconstruction model is obtained.

[0042] Exemplarily, the data reconstruction model is used to perform data reconstruction on the first window data to obtain a first reconstruction result corresponding to the first window data, and the difference between the first reconstruction result and the first window data is calculated to obtain the first error information.

[0043] Exemplarily, based on the first error information, statistical indicators such as the mean and variance of the reconstruction error of each first window data are calculated, and then based on the error statistical results, the first anomaly score of each first window data is calculated.

[0044] Exemplarily, a threshold of anomaly score is set to determine whether the first window data is abnormal. The threshold can be determined based on historical data, expert experience or statistical methods, so that data in the first operation data whose first anomaly score exceeds the threshold is marked as abnormal data to form first abnormal data.

[0045] In addition, since the first window data is the data corresponding to the target window, when the first abnormal score of the first window data exceeds the threshold, the data corresponding to the latest addition of the first window data to the target window is determined as the first abnormal data.

[0046] In some embodiments, the step of performing abnormal data identification on the first operation data according to the first abnormal score to obtain the first abnormal data includes: obtaining a target error corresponding to the training data from the data reconstruction model, and determining a corresponding abnormal segmentation threshold according to the target error; performing abnormal data identification on the first operation data according to the abnormal segmentation threshold and the first abnormal score to obtain the first abnormal data; wherein the abnormal segmentation threshold is obtained according to the following formula:

[0047]

[0048] in, represents the abnormal segmentation threshold, M represents the data volume corresponding to the training data, Represents the target error corresponding to the i-th training data.

[0049] Exemplarily, the target error corresponding to the training data under the final data reconstruction model is obtained, so as to determine the abnormal segmentation threshold using the target error according to the following formula:

[0050]

[0051] in, represents the abnormal segmentation threshold, M represents the amount of data corresponding to the training data, Represents the target error corresponding to the i-th training data.

[0052] Exemplarily, by calculating the target error of the training data and setting the abnormal segmentation threshold based on the target error, the variation range of normal training data can be more accurately reflected, and the reliance on manual experience is reduced, making the threshold setting more objective and repeatable. Thus, by calculating the target error and setting the abnormal segmentation threshold, a quantitative evaluation standard is provided, making the abnormal detection results more reliable. And as the training data is updated, the target error and abnormal segmentation threshold can be recalculated regularly, so as to continuously optimize the performance of abnormal detection, and also realize automatic threshold setting, reduce the need for manual intervention, and can quickly detect abnormalities on new data, improving processing speed.

[0053] Exemplarily, abnormal data identification is performed on the first operation data using the determined abnormal segmentation threshold and the first abnormal score. For example, the first abnormal score of each first window data is compared with the abnormal segmentation threshold. If the first abnormal score exceeds the set abnormal segmentation threshold, the corresponding data in the corresponding first operation data when it is most recently added to the target window is determined as the first abnormal data.

[0054] Step S103: Obtain a second imaging state corresponding to the second operating data from the first imaging state according to the second operating data.

[0055] Exemplarily, the second imaging state corresponding to the second operating data is obtained from the first imaging state according to the mapping relationship between the first operating data and the second operating data.

[0056] Step S104: performing feature selection on the second operation data using the neighborhood information to obtain target-related features corresponding to the injection process.

[0057] Exemplarily, a neighborhood calculation is performed on each sub-data in the second operation data to obtain the neighborhood data corresponding to each sub-data, and then the Pearson correlation coefficient, Spearman correlation coefficient, etc. are used to perform correlation analysis on the neighborhood data to calculate the correlation between each feature and the injection process, and then the features that are highly correlated with the injection process are selected as target association features.

[0058] In some embodiments, the use of neighborhood information to perform feature selection on the second operation data to obtain target-related features corresponding to the injection process includes: performing neighborhood calculation on the sub-operation data under each target parameter in the second operation data to obtain relevant neighborhood information corresponding to the sub-operation data; calculating the first entropy value information corresponding to the sub-operation data according to the relevant neighborhood information, and calculating the second entropy value information between any two target parameters according to the relevant neighborhood information; determining the neighborhood mutual information between any two target parameters according to the first entropy value information and the second entropy value information; determining the neighboring data corresponding to the target parameter according to the neighborhood mutual information, and determining the neighboring density corresponding to the target parameter according to the neighboring data; sorting the target parameters according to the neighboring density to obtain the importance corresponding to the target parameter; determining the clustering center corresponding to the second operation data when clustering the data according to the importance; performing data clustering on the second operation data according to the clustering center to obtain the target clustering result, and performing feature screening on the target parameter according to the target clustering result to obtain the target-related features corresponding to the injection process.

[0059] Exemplarily, for each sub-operation data under each target parameter in the second operation data, the neighborhood based on distance is used to find the relevant neighborhood information in its neighborhood. Thus, the local entropy of each sub-operation data is calculated according to the relevant neighborhood information, reflecting the uncertainty or complexity of the sub-operation data, and then obtaining the first entropy value information of the sub-operation data.

[0060] Exemplarily, relevant neighborhood information corresponding to any two sub-operation data under the target parameter is obtained, and then a joint entropy calculation is performed based on the two relevant neighborhood information to obtain the second entropy value information between any two target parameters, thereby reflecting the dependency relationship between the two target parameters.

[0061] Exemplarily, based on the first entropy value information and the second entropy value information, the neighborhood mutual information between any two target parameters is calculated. The neighborhood mutual information measures the degree of mutual dependence between the two target parameters and is used to evaluate the correlation between the target parameters.

[0062] Exemplarily, after sorting from high to low according to the neighborhood mutual information, the corresponding data when the neighborhood mutual information is greater than a preset value is determined as the neighbor data corresponding to the sub-operation data of each target parameter, and the mean is calculated according to the neighborhood mutual information to obtain the neighbor density corresponding to the target parameter, reflecting the correlation strength between the target parameter and other parameters.

[0063] Exemplarily, the target parameters are sorted according to the nearest neighbor density to determine the importance of each target parameter in feature selection. Thus, according to the importance of the target parameter, the corresponding cluster center is determined when the data is clustered. Generally, parameters with high importance play a more critical role in clustering. Thus, the second operation data is clustered using the determined cluster center to obtain the target clustering result, and finally, according to the target clustering result, the features strongly related to the injection process are screened out as the target associated features.

[0064] In some embodiments, determining the neighbor density corresponding to the target parameter according to the neighbor data includes: obtaining first data corresponding to the target parameter, and obtaining second data of the neighbor parameter corresponding to the target parameter from the neighbor data; calculating a first expected value corresponding to the first data and a second expected value corresponding to the second data, and calculating a third expected value corresponding to the co-occurrence of the first data and the second data; determining a target association value between the target parameter and the neighbor parameter according to the first expected value, the second expected value, and the third expected value; determining the neighbor density corresponding to the target parameter by weighted averaging the target association value and the neighborhood mutual information; wherein the target association value is obtained according to the following formula:

[0065]

[0066] in, Indicates the target parameter and the neighbor parameters The target association value between Indicates the target parameter and the neighbor parameters The third expected value of the correspondence between Indicates the target parameter The corresponding first expected value is Represents the neighbor parameter The corresponding second expected value is, Indicates the target parameter The expected value after squaring is, Represents the neighbor parameter The expected value after squaring, abs means taking the absolute value.

[0067] Exemplarily, first data corresponding to the target parameter is obtained from the second operation data, and second data corresponding to the neighbor parameter corresponding to the target parameter is obtained from the neighbor data.

[0068] Exemplarily, an expected value calculation is performed on the first data to obtain a first expected value of the first data, and an expected value calculation is performed on the second data to obtain a second expected value.

[0069] Exemplarily, the third expected value when the first data and the second data co-occur is calculated, so that by calculating different expected values, the data distribution characteristics of the target parameter and the neighboring parameter can be quantified, providing important statistical support for subsequent correlation analysis.

[0070] Exemplarily, the target correlation value between the target parameter and the neighboring parameter is determined by fusing the first expected value, the second expected value, and the third expected value according to the following formula:

[0071]

[0072] in, Indicates the target parameter and the neighbor parameters The target association value between Indicates the target parameter and the neighbor parameters The third expected value of the correspondence between Indicates the target parameter The corresponding first expected value is Represents the neighbor parameter The corresponding second expected value is, Indicates the target parameter The expected value after squaring is, Represents the neighbor parameter The expected value after squaring, abs means taking the absolute value.

[0073] Exemplarily, the target correlation value can quantify the linear correlation between the target parameter and the neighbor parameter. The closer the target correlation value is to 1, the stronger the correlation is.

[0074] Exemplarily, the target association value is determined as the target weight, so that the target association value is used to perform weighted average of the neighborhood mutual information between the neighboring parameters involved in the target parameter to determine the neighbor density corresponding to the target parameter. Thus, by combining the target association value and the neighborhood mutual information, the importance of the target parameter in the injection process can be more comprehensively evaluated, and the accuracy of feature selection can be improved.

[0075] For example, according to the calculated neighbor density, the target parameters are sorted to determine their importance in the injection process and used as the basis for feature selection. By sorting the neighbor density, the features that are strongly related to the injection process can be effectively screened out, improving the accuracy of subsequent analysis and modeling.

[0076] Specifically, the correlation between the target parameter and the neighboring parameters is accurately evaluated, and efficient feature selection is performed based on this. This process not only improves the accuracy of data processing and analysis, but also enhances the robustness and interpretability of the model, ultimately providing strong support for optimization and decision-making of the injection process.

[0077] In some embodiments, the feature screening of the target parameters according to the target clustering results to obtain the target-related features corresponding to the injection process includes: obtaining relevant parameters involved in each clustering cluster in the target clustering results, and calculating the corresponding weight information between the relevant parameters; obtaining the corresponding central parameters in the clustering cluster according to the weight information; and feature screening of the target parameters according to the central parameters to obtain the target-related features corresponding to the injection process.

[0078] Exemplarily, the relevant parameters involved in each clustering cluster are extracted from the target clustering result. These parameters are the features represented by the data points that are considered to have similar characteristics in the clustering process. The strength of the relationship between the relevant parameters is quantified based on correlation, covariance, mutual information, etc. Thus, for the relevant parameters in each clustering cluster, the weight information between them is calculated. The weight information reflects the degree of association between the relevant parameters, and the larger the value, the stronger the association.

[0079] Exemplarily, the relevant parameter corresponding to the maximum weight information is determined as the central parameter of each cluster. The central parameter represents the core feature of the cluster.

[0080] Exemplarily, the central parameter in each cluster can represent the relevant parameters in the cluster, thereby reducing parameter redundancy in subsequent data analysis and providing good support for subsequent data analysis.

[0081] Exemplarily, the central parameters corresponding to each cluster are obtained, so that all central parameters are determined as features strongly correlated with the injection process as the final target-related features.

[0082] Through the above steps, we can achieve weight analysis of relevant parameters in the clustering results, and perform feature screening based on the central parameters, so as to obtain the target-related features in the injection process. This process not only improves the accuracy of feature selection, but also simplifies the model and enhances the interpretability and practicality of the model. Ultimately, it provides reliable support for the optimization and decision-making of the injection process.

[0083] Step S105: Obtain third operation data corresponding to the target association feature from the second operation data according to the target association feature.

[0084] Exemplarily, data corresponding to the target-associated features are obtained from the second operation data according to the target-associated features to obtain the third operation data.

[0085] Step S106: determining a corresponding contrast injection state prediction model during the injection process according to the third operation data and the second contrast state.

[0086] Exemplarily, a machine learning or deep learning model, such as a random forest, a support vector machine, a neural network, etc., is determined, and the selected model is trained using the preprocessed third operation data and the second contrast state data, and the model parameters are adjusted to minimize the prediction error, thereby obtaining a contrast bolus state prediction model.

[0087] Step S107, obtaining the initial operation data corresponding to the target user during the injection process, and predicting the contrast bolus state of the initial operation data according to the contrast bolus state prediction model to obtain a target prediction result.

[0088] For example, the initial operation data corresponding to the injection process of the target patient, that is, the target user, is obtained in real time. The initial operation data is the operation data corresponding to the target user under the target-related feature, so that the contrast bolus injection state prediction model is used to predict the contrast bolus injection state of the initial operation data to obtain the target prediction result. The target prediction result is any one of the contrast is clear but the user is uncomfortable, the contrast is clear but the user is comfortable, the contrast is unclear and the user is uncomfortable, and the contrast is unclear but the user is comfortable.

[0089] In some embodiments, predicting the contrast bolus state of the initial operation data according to the contrast bolus state prediction model to obtain a target prediction result includes: performing abnormality detection on the initial operation data to obtain an abnormality detection result; adjusting the initial operation data according to the abnormality detection result to obtain target adjustment data; predicting the contrast bolus state of the target adjustment data according to the contrast bolus state prediction model to obtain the target prediction result.

[0090] Exemplarily, anomaly detection methods such as statistical methods (such as Z-score, IQR), machine learning methods (such as Isolation Forest, One-Class SVM), deep learning methods (such as Autoencoder) are used to perform anomaly detection on the initial operation data, identify abnormal points in the initial operation data, and obtain anomaly detection results.

[0091] Exemplarily, the abnormal points in the initial operation data are directly deleted according to the abnormality detection result to obtain the target adjustment data, and then the target adjustment data is predicted using the trained contrast bolus injection state prediction model to obtain the target prediction result.

[0092] In some embodiments, performing anomaly detection on the initial operation data to obtain an anomaly detection result includes: determining a target window, dividing the initial operation data according to the target window to obtain second window data; reconstructing the second window data using a data reconstruction model to obtain second error information corresponding to the second window data; determining a second anomaly score corresponding to the second window data according to the second error information; and performing anomaly detection on the second operation data according to the second anomaly score to obtain the anomaly detection result.

[0093] Exemplarily, the target window is determined. The size and range of the target window can be set according to the characteristics and requirements of the initial operation data, and can usually be determined by analyzing the time series characteristics of the data or business requirements.

[0094] Exemplarily, the initial operation data is divided according to the determined target window according to the time period to obtain the second window data, wherein the second window data includes the target data corresponding to each parameter in the initial operation data under the target window.

[0095] For example, the initial operation data includes third corresponding data corresponding to the first parameter and fourth corresponding data corresponding to the second parameter. When the target window slides in the third corresponding data, it also slides in the fourth corresponding data at the same time, so that the data corresponding to the target window simultaneously under the third corresponding data and the fourth corresponding data is determined as the second window data.

[0096] Exemplarily, normal operation data is obtained through data labeling, and then the normal operation data is compressed and reconstructed through a neural network model, and normal reconstructed data corresponding to the normal operation data is output. When the error between the normal operation data and the normal reconstructed data meets a preset condition, a data reconstruction model is obtained.

[0097] Exemplarily, the data reconstruction model is used to reconstruct the second window data to obtain a second reconstruction result corresponding to the second window data, and the difference between the second reconstruction result and the second window data is calculated to obtain the second error information.

[0098] Exemplarily, based on the second error information, statistical indicators such as the mean and variance of the reconstruction error of each second window data are calculated, and then based on the error statistical results, the second abnormality score of each second window data is calculated.

[0099] Exemplarily, a threshold of anomaly score is set to determine whether the second window data is abnormal. The threshold can be determined based on historical data, expert experience or statistical methods, so that data in the initial operation data whose second anomaly score exceeds the threshold is marked as abnormal data to form second abnormal data.

[0100] In addition, since the second window data is the corresponding data under the target window, when the second anomaly score of the second window data exceeds the threshold, the data corresponding to the time when the second window data is most recently added to the target window is determined as the second anomaly data, and then the second anomaly data is determined as the anomaly detection result.

[0101] In some embodiments, adjusting the initial operation data according to the abnormality detection result to obtain target adjustment data includes: when the abnormality detection result is an empty set, determining the initial operation data as the target adjustment data; when the abnormality detection result is not an empty set, deleting the abnormality detection result from the initial operation data to obtain the target adjustment data.

[0102] Exemplarily, when the anomaly detection result is an empty set, it means that there is no abnormal data in the initial operation data, and the initial operation data is directly determined as the target adjustment data; when the anomaly detection result is not an empty set, it means that there is data with abnormal collection in the initial operation data, and the anomaly detection result is deleted from the initial operation data to obtain the target adjustment data. The anomaly detection result is statistically analyzed to obtain the number of anomalies. When the number of anomalies is greater than a preset value, it means that there is an abnormality in the sensor that collects information from the target user, and a warning needs to be issued to adjust the device that collects the target user, so as to obtain more accurate initial operation data.

[0103] Step S108: adjusting the initial operation data according to the target prediction result to obtain the target operation data corresponding to the target user during the injection process.

[0104] For example, when the target prediction result is that the imaging is clear and the user is comfortable, there is no need to adjust the initial operation data, and the initial operation data can be directly determined as the target operation data corresponding to the target user during the injection process.

[0105] Exemplarily, when the target prediction result is any one of the following: the contrast is clear but the user is uncomfortable, the contrast is unclear and the user is uncomfortable, and the contrast is unclear but the user is comfortable, the initial operation data is adjusted until the target prediction result is that the contrast is clear and the user is comfortable, thereby obtaining the target operation data corresponding to the target user during the injection process.

[0106] See also Figure 2 , Figure 2A wireless intelligent angiography contrast injection intelligent control system 200 is provided in the embodiment of the present application. The wireless intelligent angiography contrast injection intelligent control system 200 includes a data acquisition module 201, a data processing module 202, a first screening module 203, a feature analysis module 204, a second screening module 205, a model determination module 206, a data prediction module 207, and a data adjustment module 208, wherein the data acquisition module 201 is used to obtain the first operation data corresponding to the historical user during the injection process and the first contrast state corresponding to the first operation data; the data processing module 202 is used to identify the abnormal data of the first operation data to obtain the first abnormal data, and process the first operation data according to the first abnormal data to obtain the second operation data; the first screening module 203 is used to obtain the second operation data from the first contrast state according to the second operation data a second contrast state corresponding to the second operation data; a feature analysis module 204, used for performing feature selection on the second operation data using neighborhood information to obtain target-related features corresponding to the injection process; a second screening module 205, used for obtaining third operation data corresponding to the target-related features from the second operation data according to the target-related features; a model determination module 206, a user determines a contrast push state prediction model corresponding to the injection process according to the third operation data and the second contrast state; a data prediction module 207, used for obtaining initial operation data corresponding to the target user during the injection process, and predicting the contrast push state of the initial operation data according to the contrast push state prediction model to obtain a target prediction result; a data adjustment module 208, used for adjusting the initial operation data according to the target prediction result to obtain the target operation data corresponding to the target user during the injection process.

[0107] In some implementations, during the process of identifying abnormal data on the first operation data to obtain first abnormal data, the data processing module 202 executes:

[0108] Determine a target window, and divide the first operation data according to the target window to obtain first window data;

[0109] Reconstructing the first window data using a data reconstruction model to obtain first error information corresponding to the first window data;

[0110] Determine a first abnormality score corresponding to the first window data according to the first error information;

[0111] Abnormal data identification is performed on the first operation data according to the first abnormal score to obtain the first abnormal data.

[0112] In some implementations, during the process of identifying abnormal data on the first operation data according to the first abnormal score to obtain the first abnormal data, the data processing module 202 executes:

[0113] Obtaining a target error corresponding to the training data from the data reconstruction model, and determining a corresponding abnormal segmentation threshold according to the target error;

[0114] Performing abnormal data identification on the first operation data according to the abnormal segmentation threshold and the first abnormal score to obtain the first abnormal data;

[0115] The abnormal segmentation threshold is obtained according to the following formula:

[0116]

[0117] in, represents the abnormal segmentation threshold, M represents the data volume corresponding to the training data, Represents the target error corresponding to the i-th training data.

[0118] In some implementations, the feature analysis module 204 performs, during the process of performing feature selection on the second operation data using the neighborhood information to obtain the target associated features corresponding to the injection process:

[0119] Performing neighborhood calculation on the sub-operation data under each target parameter in the second operation data to obtain relevant neighborhood information corresponding to the sub-operation data;

[0120] Calculating first entropy value information corresponding to the sub-operation data according to the relevant neighborhood information, and calculating second entropy value information between any two of the target parameters according to the relevant neighborhood information;

[0121] Determine the neighborhood mutual information between any two of the target parameters according to the first entropy value information and the second entropy value information;

[0122] Determine neighbor data corresponding to the target parameter according to the neighborhood mutual information, and determine neighbor density corresponding to the target parameter according to the neighbor data;

[0123] Sort the target parameters according to the neighbor density to obtain the importance of the target parameters;

[0124] Determining, according to the importance, a cluster center corresponding to the second operation data when data clustering is performed;

[0125] The second operation data is clustered according to the cluster center to obtain a target clustering result, and the target parameters are feature screened according to the target clustering result to obtain the target association feature corresponding to the injection process.

[0126] In some implementations, the feature analysis module 204, in the process of determining the neighbor density corresponding to the target parameter according to the neighbor data, performs:

[0127] Obtaining first data corresponding to the target parameter, and obtaining second data of a neighbor parameter corresponding to the target parameter from the neighbor data;

[0128] Calculating a first expected value corresponding to the first data and a second expected value corresponding to the second data, and calculating a third expected value corresponding to when the first data and the second data appear together;

[0129] Determine a target correlation value between the target parameter and the neighbor parameter according to the first expected value, the second expected value, and the third expected value;

[0130] Determine the neighbor density corresponding to the target parameter by performing weighted averaging based on the target association value and the neighborhood mutual information;

[0131] The target correlation value is obtained according to the following formula:

[0132]

[0133] in, Indicates the target parameter and the neighbor parameters The target association value between Indicates the target parameter and the neighbor parameters The third expected value of the correspondence between Indicates the target parameter The corresponding first expected value is Represents the neighbor parameter The corresponding second expected value is, Indicates the target parameter The expected value after squaring is, Represents the neighbor parameter The expected value after squaring, abs means taking the absolute value.

[0134] In some embodiments, the feature analysis module 204 performs, during the process of performing feature screening on the target parameters according to the target clustering result to obtain the target associated features corresponding to the injection process:

[0135] Obtain relevant parameters involved in each clustering cluster in the target clustering result, and calculate corresponding weight information between the relevant parameters;

[0136] Obtaining the corresponding center parameter in the cluster according to the weight information;

[0137] The target parameters are feature screened according to the central parameters to obtain the target-related features corresponding to the injection process.

[0138] In some embodiments, the data prediction module 207 performs, during the process of predicting the contrast bolus state of the initial operation data according to the contrast bolus state prediction model to obtain a target prediction result:

[0139] Performing anomaly detection on the initial operation data to obtain an anomaly detection result;

[0140] Adjusting the initial operation data according to the abnormality detection result to obtain target adjustment data;

[0141] The contrast bolus state of the target adjustment data is predicted according to the contrast bolus state prediction model to obtain the target prediction result.

[0142] In some implementations, the data prediction module 207, during the process of performing anomaly detection on the initial operation data to obtain an anomaly detection result, performs:

[0143] Determine a target window, and divide the initial operation data according to the target window to obtain second window data;

[0144] Reconstructing the second window data using a data reconstruction model to obtain second error information corresponding to the second window data;

[0145] Determine a second abnormality score corresponding to the second window data according to the second error information;

[0146] Anomaly detection is performed on the second operation data according to the second anomaly score to obtain the anomaly detection result.

[0147] In some implementations, the data prediction module 207, during the process of adjusting the initial operation data according to the abnormality detection result to obtain the target adjustment data, performs:

[0148] When the abnormality detection result is an empty set, the initial operation data is determined as the target adjustment data;

[0149] When the abnormality detection result is not an empty set, the abnormality detection result is deleted from the initial operation data to obtain the target adjustment data.

[0150] In some embodiments, the wireless intelligent angiography imaging push injection intelligent control system 200 can be applied to a terminal device.

[0151] It should be noted that, technical personnel in the relevant field can clearly understand that, for the convenience and simplicity of description, the specific working process of the wireless intelligent angiography contrast injection intelligent control system 200 described above can refer to the corresponding process in the aforementioned wireless intelligent angiography contrast injection intelligent control method embodiment, and will not be repeated here.

[0152] An embodiment of the present invention also provides a storage medium for computer-readable storage, wherein the storage medium stores one or more programs, and the one or more programs can be executed by one or more processors to implement the steps of any one of the wireless intelligent angiography instrument angiography push injection intelligent control methods provided in the description of the embodiment of the present invention.

[0153] The storage medium may be an internal storage unit of the terminal device in the aforementioned embodiment, such as a hard disk or memory of the terminal device. The storage medium may also be an external storage device of the terminal device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the terminal device.

[0154] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware embodiment, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or implemented as hardware, or implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or non-transitory medium) and a communication medium (or transient medium). As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules, or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. In addition, it is well known to those skilled in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0155] It should be understood that the term "and / or" used in the present specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, including these combinations. It should be noted that, in this article, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "including a..." does not exclude the presence of other identical elements in the process, method, article or system including the element.

[0156] The serial numbers of the embodiments of the present invention are only for description and do not represent the advantages and disadvantages of the embodiments. The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A wireless intelligent angiography instrument imaging push injection intelligent control system, characterized in that: include: A data acquisition module, used for obtaining first operation data corresponding to the historical user during the injection process and a first contrast state corresponding to the first operation data; a data processing module, configured to perform abnormal data identification on the first operation data to obtain first abnormal data, and process the first operation data according to the first abnormal data to obtain second operation data; a first screening module, configured to obtain a second imaging state corresponding to the second operation data from the first imaging state according to the second operation data; A feature analysis module, used for performing feature selection on the second operation data using neighborhood information to obtain target-related features corresponding to the injection process, including: performing neighborhood calculation on the sub-operation data under each target parameter in the second operation data to obtain relevant neighborhood information corresponding to the sub-operation data; calculating first entropy value information corresponding to the sub-operation data according to the relevant neighborhood information, and calculating second entropy value information between any two target parameters according to the relevant neighborhood information; determining neighborhood mutual information between any two target parameters according to the first entropy value information and the second entropy value information; determining neighbor data corresponding to the target parameter according to the neighborhood mutual information, and determining neighbor density corresponding to the target parameter according to the neighbor data, including: obtaining data corresponding to the target parameter, and obtaining second data of the neighbor parameter corresponding to the target parameter from the neighbor data; calculating a first expected value corresponding to the first data and a second expected value corresponding to the second data, and calculating a third expected value corresponding to the common appearance between the first data and the second data; determining a target association value between the target parameter and the neighbor parameter according to the first expected value, the second expected value and the third expected value; determining the neighbor density corresponding to the target parameter by weighted average based on the target association value and the neighborhood mutual information; Among them, the target association value is obtained according to the following formula: ; in, Indicates the target parameter and neighbor parameters The target correlation value between Indicates the target parameter and neighbor parameters The corresponding third expected value between Indicates the target parameter The corresponding first expected value is Represents the neighbor parameter The corresponding second expected value is, Indicates the target parameter The expected value after squaring is, Represents the neighbor parameter The expected value after squaring, abs means taking the absolute value; sorting the target parameters according to the nearest neighbor density to obtain the importance of the target parameters; determining the corresponding cluster center when clustering the second operation data according to the importance; clustering the second operation data according to the cluster center to obtain the target clustering result, and performing feature screening on the target parameters according to the target clustering result to obtain the corresponding target correlation feature in the injection process; A second screening module, configured to obtain third operation data corresponding to the target association feature from the second operation data according to the target association feature; A model determination module, used for determining a corresponding contrast injection state prediction model during the injection process according to the third operation data and the second contrast state; A data prediction module, used for obtaining the initial operation data corresponding to the target user during the injection process, and predicting the contrast bolus injection state of the initial operation data according to the contrast bolus injection state prediction model to obtain a target prediction result; The data adjustment module is used to adjust the initial operation data according to the target prediction result to obtain the target operation data corresponding to the target user during the injection process.

2. The system according to claim 1, characterized in that The performing abnormal data identification on the first operation data to obtain first abnormal data includes: Determine a target window, and divide the first operation data according to the target window to obtain first window data; Reconstructing the first window data using a data reconstruction model to obtain first error information corresponding to the first window data; Determine a first abnormality score corresponding to the first window data according to the first error information; Abnormal data identification is performed on the first operation data according to the first abnormal score to obtain the first abnormal data.

3. The system according to claim 2, characterized in that The performing abnormal data identification on the first operation data according to the first abnormal score to obtain the first abnormal data includes: Obtaining a target error corresponding to the training data from the data reconstruction model, and determining a corresponding abnormal segmentation threshold according to the target error; Performing abnormal data identification on the first operation data according to the abnormal segmentation threshold and the first abnormal score to obtain the first abnormal data; The abnormal segmentation threshold is obtained according to the following formula: ; in, represents the abnormal segmentation threshold, M represents the data volume corresponding to the training data, Represents the target error corresponding to the i-th training data.

4. The system according to claim 1, characterized in that The method of performing feature screening on target parameters according to the target clustering result to obtain target-related features corresponding to the injection process includes: Obtain relevant parameters involved in each clustering cluster in the target clustering result, and calculate corresponding weight information between the relevant parameters; Obtaining the corresponding center parameter in the cluster according to the weight information; The target parameters are feature screened according to the central parameters to obtain the target-related features corresponding to the injection process.

5. The system according to claim 1, characterized in that The step of predicting the contrast bolus state of the initial operation data according to the contrast bolus state prediction model to obtain a target prediction result includes: Performing anomaly detection on the initial operation data to obtain an anomaly detection result; Adjusting the initial operation data according to the abnormality detection result to obtain target adjustment data; The contrast bolus state of the target adjustment data is predicted according to the contrast bolus state prediction model to obtain the target prediction result.

6. The system according to claim 5, characterized in that The performing anomaly detection on the initial operation data to obtain an anomaly detection result includes: Determine a target window, and divide the initial operation data according to the target window to obtain second window data; Reconstructing the second window data using a data reconstruction model to obtain second error information corresponding to the second window data; Determine a second abnormality score corresponding to the second window data according to the second error information; Anomaly detection is performed on the second operation data according to the second anomaly score to obtain the anomaly detection result.

7. The system according to claim 6, characterized in that The adjusting the initial operation data according to the abnormality detection result to obtain target adjustment data includes: When the abnormality detection result is an empty set, the initial operation data is determined as the target adjustment data; When the abnormality detection result is not an empty set, the abnormality detection result is deleted from the initial operation data to obtain the target adjustment data.

Images