Design And Development Of Screening System For Congental Heart Disease

Congenital heart disease(CHD)is a high incidence disease,but the clinical diagnosis and screening methods are limited by the high cost of ultrasound diagnosis and doctor's auscultation experience.At present,many designed heart sound screening devices still depend on computer,Android and other terminal systems.So,a novel screening system,which has the ability of working independently and can collect and analyze heart sound data at any time,has high engineering significance and economic value.Such system can also help solve the problem of limited medical resource in remote areas.Under the support of three-year action plan of promoting clinical skills and clinical innovation in Shanghai Shenkang Medical Center,"Research and Optimization of Intelligent Expert Assistant Screening System for Congenital Heart Disease in Children with the Technology of Cloud Based Big Data"(16CR3079B)and the cooperation with Xinhua Hospital affiliated to School of Medicine of Shanghai Jiao Tong University,the purpose of this paper is to study the design method of computer-aided medical equipment and explore the technology of diagnosis with cloud big data.With the design of screening system for congenital heart disease as research object,research is carried out in three aspects:(1)Based on the design method: requirement analysis,function synthesis,concept selection,the concept scheme of screening system is constructed.According to the concept scheme,the detailed design of hardware system is implemented: The auscultation head that meets the requirements of acoustics and ergonomics;The selection of sensor and the design and simulation of preprocessing circuit that is suitable for weak heart sound signal with low frequency;The design of display and storage function which meets the interactive requirement;The design of the integrated system box.With WIFI connected and the heart sound analysis function provided by the cloud server,user only need to operate the screening system itself to collect heart sound data and get screening result.They do not need to operate computers,Android and other auxiliary devices.(2)In the preprocessing of heart sounds,a method of heart sound segmentation based on multi-auscultation area is proposed,many features are extracted to establish models,a ensembled random forest model based on multi-auscultation areas is established.In this paper,an unsegmented heart sound analysis method,based on convolution neural network and connectionist temporal classification method is established.This method avoids the difficulty of segmenting heart sound with strong murmur.The model shows a good performance in the classification of heart sounds when the train-data and test-data is not split.The above analysis method can provide powerful guidance for the classification of heart sounds under big data.(3)The software of heart sound screening system is established.The software of screening system includes the client in the hardware system and the server program in the cloud server.The client can select data from cloud database and update other information.The model based on TensorFlow is built by using the designed analysis method with python.Such software connects the hardware and analysis method in the cloud server.Based on the hardware design,algorithm analysis and software design,the design and development of screening system for congenital heart disease are completed in this paper.This design provides a solution to the problem of low prevalence of screening for congenital heart disease,explores the design method and technology of computer-aided medical equipment and provides a reference for the computer-aided diagnosis and treatment of other diseases.