Lung disease auscultation system, signal processing method and equipment based on convolutional neural network

Abstract

The present invention proposes a pulmonary disease auscultation system, signal processing method and equipment based on a convolutional neural network. The system includes: a preprocessing module configured to sequentially normalize the acquired breath sound signals; a data conversion module configured to The audio feature is extracted from the normalized data to generate Mel spectrum data; the data enhancement module is configured to use the data enhancement model to amplify the generated Mel spectrum data; the convolutional neural network module is configured as The trained deep learning model is used to extract features from the amplified mel spectrum data; the feature classification module is configured to classify the extracted features using the trained low-discrepancy forest classifier to obtain classification results. It can achieve accurate feature classification for small samples, and the system has a small amount of calculation. It can be deployed on embedded computer equipment and can realize remote diagnosis and treatment.

Description

technical field [0001] The present invention relates to the relevant technical field of intelligent medical information technology, in particular to a pulmonary auscultation system based on a convolutional neural network, a signal processing method and equipment. Background technique [0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art. [0003] Respiratory diseases are a common problem worldwide. Mild cases manifest as cough, chest pain, and affected breathing, and severe cases manifest as dyspnea, hypoxia, and even death from respiratory failure. Smoking is the most common cause of respiratory disease, but it is sometimes caused by genetic and environmental factors. [0004] Automatic classification of breath sounds has the potential to detect abnormalities in the early stages of respiratory dysfunction, thereby improving the effectiveness of decision-making. At present...

AI Technical Summary

Problems solved by technology

Although the former is accurate and easy to judge, it requires relevant professional equipment; although the latter is low in cost, it requires a higher level of experience for doctors, and many areas cannot be equipped with professional equipment or experienced doctors in time. The correct diagnosis of the disease has a greater impact

[0005] Some R&D personnel have begun to use deep learning models for auxiliary diagnosis. Some of the existing models have very complex algorithms and huge network structures. It is very difficult to deploy them on mobile terminals, and they cannot be applied on low-cost computers, so they cannot benefit the vast underdeveloped areas.

Images