Abnormal electrocardiogram recognition method based on ultra-complete characteristics

Abstract

The invention relates to an abnormal cardioelectric recognize method based on ultra-complete character. Wherein. it first uses secondary sample wavelet to process R-point check on the cardioelectric data, segments and pretreats the cardioelectric data based on the R-point position; then uses independent component method and disperse wavelet convert method to extract one ultra-complete character group from each heart beat, and contract the character via communication method; at last uses the vector method to train the extracted character to obtain one vector mode, to automatically recognize and classify the new heart beat data.

Description

technical field [0001] The invention relates to a method in the technical field of signal processing, in particular to a method for identifying abnormal ECG based on overcomplete features. Background technique [0002] Electrocardiogram (ECG) is often used to detect heart activity in clinical medicine and has very important clinical value. Record a complete ECG waveform on the recording paper of the electrocardiograph, mainly including: P wave (the P wave that first appears in each wave of the ECG is the P wave representing the excitation process of the left and right atria), QRS wave group (representing the propagation process of the excitation of the two ventricles) A typical QRS complex includes three connected fluctuations, the first downward wave is the Q wave, followed by the Q wave, a narrow and high upward wave is the R wave, and another connected to the R wave is the R wave. The lower wave is the S wave), the S-T segment (the flat line from the end of the QRS wave ...

AI Technical Summary

Problems solved by technology

Its shortcoming is that it can only achieve good results for normal heartbeats, the accuracy of the other four types is not high, and it only classifies 5 types of ECG signals. Secondly, due to the large number of components used, the computational complexity of the training phase is very high. , if the dimension of the feature vector used is too high, it will affect the performance of recognition, so that real-time ECG automatic recognition cannot be achieved

[0004] Usually, a real-valued signal can be represented by a linear combination of a set of basis functions, which is an effective method for encoding high-dimensional data space. For example, Fourier or wavelet can provide an effective representation of the signal, but they Underdetermined signal cannot be clearly indicated

Images