Basic heart sound recognition method and device

Abstract

The invention relates to a heart sound recognition method. The method comprises the following steps of collecting a plurality of original heart sound data; performing low-pass filtering on the plurality of original heart sound data, and calculating homomorphic envelopes corresponding to the plurality of original heart sound data; screening the plurality of homomorphic envelopes by adopting a double-threshold method to obtain at least one alternative heart sound data segment; performing short-time Fourier transform on the at least one alternative heart sound data segment to obtain an alternative heart sound segment time-frequency spectrum corresponding to the at least one alternative heart sound data segment; and inputting the at least one alternative heart sound time-frequency spectrum into a heart sound recognition branch convolutional neural network for classification to obtain a classification result of the at least one alternative heart sound time-frequency spectrum.

Description

technical field [0001] The present invention relates to the field of neural networks, in particular to a basic heart sound recognition method and device based on a branched convolutional neural network. Background technique [0002] Heart valve disease is a relatively common cardiovascular disease at present, and the mainstream cardiac examination methods mainly include cardiac CT, cardiac magnetic resonance and cardiac ultrasound. However, the inspection process of the above methods requires the participation of professionals, and is not suitable for promotion in the home environment. [0003] As a physiological signal of the human body, heart sounds can reflect the health of the heart valves. The use of heart sounds to diagnose heart valve diseases is cheap and easy to popularize. [0004] Compared with traditional mechanical stethoscopes, electronic stethoscopes can save heart sound data in digital form and can perform automatic diagnosis at the same time, so the applic...

AI Technical Summary

Problems solved by technology

However, traditional S1 and S2 recognition is based on clustering, and the accuracy is relatively low

At the same time, the processing flow of some current identification methods for S1 and S2 is relatively complicated, and they are all binary classifications of S1 and S2.

For the binary classification method, if the heart sound samples also include occasional environmental noises, the noises will be wrongly identified as S1 or S2, resulting in errors in subsequent heart sound segmentation

Obviously, the current technology is not conducive to real-time recognition, and the noise in the heart sound will also be misidentified

Images