Electrocardiogram analysis method and device based on target detection

Abstract

The invention provides an electrocardiogram analysis method based on target detection, which comprises the following steps of: constructing a target detection model by using collected sample cardiac beat data, inputting an electrocardiogram signal to be analyzed into the target detection model, and collecting an analysis result, the step for constructing the target detection model comprises the following steps: converting collected sample heart beat data into a two-dimensional image, performing preliminary feature extraction on the two-dimensional image, and outputting a feature map of an original image; fusing and extracting the feature map of the original image, and outputting a final feature map; performing operation on the final feature map to complete regression and classification, and outputting a plurality of groups of QRS complex wave positions, heart beat categories and confidence coefficients; screening the output QRS complex wave position, heart beat category and confidence, and outputting an analysis result; according to the method provided by the invention, the problems that the features are manually extracted and the features are influenced by R peak detection precision are relieved, the QRS complex waves do not need to be manually detected, and the extracted features can also be used as model input and an analysis result can be obtained.

Description

technical field [0001] The invention belongs to the technical field of automatic detection of electrocardiogram signals, in particular to an electrocardiogram analysis method and device based on target detection. Background technique [0002] The electrocardiogram serves as a record of heart activity and provides important information about the state of the heart. Analyzing the electrocardiogram is a necessary means for early diagnosis of heart disease patients. On the one hand, it is difficult for doctors to analyze the long-recorded ECG within a limited time; on the other hand, without tool support, it is almost impossible for people to recognize the morphological changes of the ECG signal. Therefore, an effective computer-aided diagnosis system is needed to solve this problem. [0003] The current ECG analysis methods mainly include methods based on traditional machine learning and deep learning methods based on convolutional neural networks. [0004] The method based ...

AI Technical Summary

Problems solved by technology

[0006] a pattern such as figure 2 As shown, the position of the R peak is located by the QRS complex wave detector to divide the heart beat, and then no manual design is required to extract the features, which are directly fed to the CNN. Using the CNN as an automatic feature extractor, the data features will be automatically learned by the network during the training process. , but this mode still depends on the detection accuracy of the previous QRS complex

[0007] Another pattern such as image 3 As shown, feeding a continuous ECG signal data to the network can perform abnormal interpretation without QRS complex wave detection and manual design and extraction of features, but the frequency and location of abnormal heart beats cannot be obtained from the results through this mode Wait for more information, not conducive to visualization and further analysis

In addition, many high-accuracy methods choose to use multi-lead ECG signals in order to obtain more features. On the one hand, the parameters of the network model are increased, and the computational complexity is increased, which is not conducive to real-time detection. On the other hand, if it is implemented in products, Need to collect signals from multiple positions of the user's body, reducing user comfort

Images