阅读说明：本技术 一种基于深度学习的智能听诊系统 (Intelligent auscultation system based on deep learning ) 是由 陈伟 徐晨 解刚才 吴辉群 于 2019-10-12 设计创作，主要内容包括：本发明公开了一种基于深度学习的智能听诊系统,包括用户端智能设备、远程数据库管理系统以及远程服务器诊断系统；用户端智能设备包括心音高灵敏度传感器,用户端高灵敏度传感器采集的生理数据通过蓝牙或者WIFI与用户端手机通信；用户端手机上设有智能诊断APP,包括通信模块、用户认证模块、信号智能判断模块、深度学习诊断模块；所述的远程数据库管理系统包括认证模块、心音数据模块,心音数据模块包括标准数据模块、用户数据模块；远程服务器诊断系统服务器上部署深度学习模型,深度学习模型包括训练模块、通信信令模块。本发明可高精度地预测心音是否异常,克服了现有智能听诊器诊断费用高,等待时间长的问题。(The invention discloses an intelligent auscultation system based on deep learning, which comprises user-side intelligent equipment, a remote database management system and a remote server diagnosis system; the intelligent equipment of the user side comprises a heart sound high-sensitivity sensor, and physiological data acquired by the user side high-sensitivity sensor is communicated with a user side mobile phone through Bluetooth or WIFI; the intelligent diagnosis APP is arranged on the user side mobile phone and comprises a communication module, a user authentication module, a signal intelligent judgment module and a deep learning diagnosis module; the remote database management system comprises an authentication module and a heart sound data module, wherein the heart sound data module comprises a standard data module and a user data module; the remote server diagnosis system server is provided with a deep learning model, and the deep learning model comprises a training module and a communication signaling module. The invention can predict whether the heart sound is abnormal or not with high precision, and overcomes the problems of high diagnosis cost and long waiting time of the existing intelligent stethoscope.)

1. The utility model provides an intelligence auscultation system based on degree of depth study which characterized in that: the system comprises user-side intelligent equipment, a remote database management system and a remote server diagnosis system;

the intelligent equipment at the user side comprises a heart sound high-sensitivity sensor, and physiological data acquired by the user side high-sensitivity sensor is communicated with a user side mobile phone through Bluetooth or WIFI; the intelligent diagnosis APP is arranged on the user side mobile phone, and the functions of the intelligent diagnosis APP comprise a Bluetooth communication module, a WIFI communication module, a user information module, a signal intelligent judgment module and a deep learning diagnosis module; the remote database management system comprises an authentication module, a communication module and a heart sound data module, wherein the heart sound data module comprises a standard data module and a user information management module;

the remote server diagnosis system server is provided with a deep learning model, and the deep learning model comprises a training module and a communication signaling module.

2. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the heart sound high-sensitivity sensor picks up human heart sound signals and uploads the human heart sound signals to the user side mobile phone through Bluetooth or WIFI, and local analysis and diagnosis are carried out through an intelligent diagnosis APP on the user side mobile phone.

3. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the Bluetooth communication module is communicated with a Bluetooth chip of the high-precision sensor through a mobile phone Bluetooth protocol stack and Bluetooth hardware, and transmits the heart sound data acquired by the sensor to a mobile phone of a user side.

4. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the WIFI communication module uploads local user information and collected user heart sound data to a remote database management system, and meanwhile, the WIFI communication module is in charge of communicating with a remote server diagnosis system to update an intelligent diagnosis APP on a user side mobile phone.

5. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the user information module records information such as age, sex, physical condition, hypertension family and the like of the user.

6. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the intelligent signal judgment module and the deep learning diagnosis module perform recognition processing analysis on the acquired heart sound data, give diagnosis results and upload the diagnosis results to a remote database management system.

7. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the deep learning algorithm in the deep learning diagnosis module can diagnose whether the heart sound data of the user is abnormal or not with high precision, and the disease type of the heart sound is predicted by using the deep learning model.

8. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the remote database management system selects MySQL as a database management system and is erected in a remote data server; the remote database management system comprises an authentication module, a communication module and a heart sound data module, wherein the heart sound data module comprises a standard data module and a user information management module; the authentication module is responsible for the identity recognition password verification function of the user and the administrator, and the user authenticated by the administrator can obtain user information and corresponding cardiopulmonary sound data; the communication module is mainly responsible for communicating with the smart phone and the remote server diagnosis system through the Internet; the user information management module records the age, sex, physical condition and hypertension family information of the user; the standard data module is responsible for managing training data for training deep learning, and the training data comprises public labeled heart sound data and expert labeled user data.

9. The intelligent auscultation system based on deep learning of claim 1, characterized in that: the high-sensitivity sensor at the user end can suppress background noise to the maximum extent and improve the signal-to-noise ratio by utilizing the high-sensitivity acoustoelectric sensor, and is responsible for converting heart sound and audio signals of a human body into electric signals, the electric signals are converted into digital audio through A/D (analog to digital), and the sampling frequency of an A/D converter is set to be 2000 Hz; then, noise and interference signals are suppressed by a band-pass Butterworth filter (25-900Hz), the amplified signals are used for a multi-channel sound encoder to encode digital audio in a wav format, and finally the digital audio in the wav format is sent to a smart phone of a user end through a wireless communication circuit.

Technical Field

The invention belongs to the field of medical diagnosis, and particularly relates to an intelligent auscultation system based on deep learning.

Background

With the development of electronic technology, various digital stethoscopes appear on the market, but the stethoscopes simply collect, record, amplify and process and analyze physiological signals. However, the intelligent stethoscope product is scarce, and the domestic market currently has only the Lobob radish intelligent stethoscope developed by Chengdu radish technology corporation and the nine-heart electronic stethoscope produced by Changhong radish corporation. The intelligent diagnosis of the two products uploads local physiological audio data to a cloud server by means of the Internet, then the online doctor diagnoses the physiological audio data, and finally the diagnosis result is fed back to the user through the network. The intelligent stethoscope has the following disadvantages: firstly, when the network is poor or the cloud access amount is increased, a large delay is caused, so that the user experience is greatly reduced. Secondly, the server-side diagnosis-based intelligent stethoscope sends locally acquired heart sound data to a cloud server for auscultation by an online doctor, answers online, and the diagnosis result depends on the diagnosis level of the online doctor, and the qualification and clinical work experience of the online doctor cannot be verified, so that the reliability of the diagnosis result is low; finally, online inquiry charges are high, which brings great economic cost to users.

Disclosure of Invention

The purpose of the invention is as follows: in view of the above, the present disclosure provides an intelligent auscultation system based on deep learning, which can implement local real-time intelligent diagnosis of physiological heart sounds of a user.

The technical scheme is as follows: in order to achieve the above object, an embodiment of the present invention adopts the following technical solutions:

intelligent auscultation system based on degree of depth study, its characterized in that: the system comprises user-side intelligent equipment, a remote database management system and a remote server diagnosis system;

the intelligent equipment at the user side comprises a heart sound high-sensitivity sensor, and physiological data acquired by the user side high-sensitivity sensor is communicated with a user side mobile phone through Bluetooth or WIFI; the intelligent diagnosis APP is arranged on the user side mobile phone, and the functions of the intelligent diagnosis APP comprise a Bluetooth communication module, a WIFI communication module, a user information module, a signal intelligent judgment module and a deep learning diagnosis module; the remote database management system comprises an authentication module, a communication module and a heart sound data module, wherein the heart sound data module comprises a standard data module and a user information management module;

the remote server diagnosis system server is provided with a deep learning model, and the deep learning model comprises a training module and a communication signaling module.

As an optimization: the heart sound high-sensitivity sensor picks up human heart sound signals and uploads the human heart sound signals to the user side mobile phone through Bluetooth or WIFI, and local analysis and diagnosis are carried out through an intelligent diagnosis APP on the user side mobile phone.

As an optimization: the Bluetooth communication module is communicated with a Bluetooth chip of the high-precision sensor through a mobile phone Bluetooth protocol stack and Bluetooth hardware, and transmits the heart sound data acquired by the sensor to a mobile phone of a user side.

As an optimization: the WIFI communication module uploads local user information and collected user heart sound data to a remote database management system, and meanwhile, the WIFI communication module is in charge of communicating with a remote server diagnosis system to update an intelligent diagnosis APP on a user side mobile phone.

As an optimization: the user information module records information such as age, sex, physical condition, hypertension family and the like of the user.

As an optimization: the intelligent signal judgment module and the deep learning diagnosis module perform recognition processing analysis on the acquired heart sound data, give diagnosis results and upload the diagnosis results to a remote database management system.

As an optimization: the deep learning algorithm in the deep learning diagnosis module can diagnose whether the heart sound data of the user is abnormal or not with high precision, and the disease type of the heart sound is predicted by using the deep learning model.

As an optimization: the remote database management system selects MySQL as a database management system and is erected in a remote data server; the remote database management system comprises an authentication module, a communication module and a heart sound data module, wherein the heart sound data module comprises a standard data module and a user information management module; the authentication module is responsible for the identity recognition password verification function of the user and the administrator, and the user authenticated by the administrator can obtain user information and corresponding cardiopulmonary sound data; the communication module is mainly responsible for communicating with the smart phone and the remote server diagnosis system through the Internet; the user information management module records the age, sex, physical condition and hypertension family information of the user; the standard data module is responsible for managing training data for training deep learning, and the training data comprises public labeled heart sound data and expert labeled user data.

As an optimization: the high-sensitivity sensor at the user end can suppress background noise to the maximum extent and improve the signal-to-noise ratio by utilizing the high-sensitivity acoustoelectric sensor, and is responsible for converting heart sound and audio signals of a human body into electric signals, the electric signals are converted into digital audio through A/D (analog to digital), and the sampling frequency of an A/D converter is set to be 2000 Hz; then, noise and interference signals are suppressed by a band-pass Butterworth filter (25-900Hz), the amplified signals are used for a multi-channel sound encoder to encode digital audio in a wav format, and finally the digital audio in the wav format is sent to a smart phone of a user end through a wireless communication circuit.

Has the advantages that: the system of the invention has the following beneficial effects:

1. the invention utilizes the intelligent equipment of the user end to make the physiological data of the user as follows: the heart sound signals are digitized and then transmitted to the smart phone of the user through Bluetooth. The mobile phone APP integrated high-precision deep learning model predicts the disease type of the mobile phone APP at the client, provides auxiliary diagnosis for the user, and brings great convenience to the user.

2. According to the invention, the mobile phone APP uploads the heart sound data and the user information of the user to the cloud database, so that abundant and valuable heart sound data are provided for clinic.

3. The remote server diagnosis system regularly retrains the deep learning model by using valuable data in the cloud database so as to improve the accuracy of model diagnosis, and updates the user side APP through the communication module.

Drawings

FIG. 1 is a schematic diagram of one embodiment of an intelligent auscultation system based on deep learning of the present invention;

fig. 2 is a schematic diagram of an embodiment of a user-side smart device APP according to the present invention;

FIG. 3 is a diagram illustrating an embodiment of MFSC feature map extraction for training a deep learning model according to the invention;

FIG. 4 is a schematic diagram of one embodiment of a deep learning based cardiopulmonary sound diagnostic process of the present invention;

FIG. 5 is a diagram illustrating the accuracy of the deep learning model of the present invention on a training set;

FIG. 6 is a diagram illustrating the accuracy of the deep learning model of the present invention on a verification set.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below so that those skilled in the art can better understand the advantages and features of the present invention, and thus the scope of the present invention will be more clearly defined. The embodiments described herein are only a few embodiments of the present invention, rather than all embodiments, and all other embodiments that can be derived by one of ordinary skill in the art without inventive faculty based on the embodiments described herein are intended to fall within the scope of the present invention.