Automatic Classification Of Fetal Heart Rate Based On Convolutional Neural Network And The Effect Of Music On Fetal Heart Rate

The government of China has been developing electronic fetal monitoring(EFM)systems to ensure the health of pregnant women and fetuses.In the EFM,it is extremely critical to identify abnormalities in the fetal heart rate(FHR).At present,the identification of fetal heart rate in China is mainly based on doctors,supplemented by computer statistical methods.Obviously,the recognition of fetal heart rate is not automated and accurate.The main purpose of this paper is to establish a more complete electronic fetal monitoring system.And then we use machine learning algorithms to improve the accuracy of identifying the fetal heart rate.In addition,we invite pregnant women volunteers to participate in experiments on the effect of music on the FHR.Therefore,this paper mainly includes the following three parts:(1)In the background of internet medical treatment,the electronic fetal monitoring system is established in this paper.The main service object of the system is mother and baby,and its services are mainly fetal monitoring.The system is based on cloud computing,including three hosts,IOS,Android,and Web.The system has been commercialized and has two hundred thousand users.(2)FHR is very significant to evaluate the status of fetus.However,based on traditional classification criteria is not accurate.With the rapid development of computer information technology,computer technology is vital for the analysis of fetal heart rate in the EFM.FHR is divided into three classes as normal,suspicious and abnormal.Through the cooperation with the hospital,we got 4473 records,including 3012 normal,1024 suspicious,437 abnormal records by our EFM system.In order to improve the accuracy of fetal status assessment,a data processing model of fetal heart rate based on convolution neural network is proposed in this paper.First,the model method divides high one-dimensional FHR records into ten d-window segments,and then use convolutional neural network(CNN)to process the FHR data in parallel.Finally,we use the voting method to determine the class of fetal heart rate records.We also made a comparative experiment,the feature extraction method based on basic statistics is used to extract the features of FHR.And then the features were applied as the input to Support Vector Machine(SVM)and Multi-layer Perceptron(MLP)to classify.According to the results of the experiment,the accuracy of classification of SVM,MLP and CNN are 79.66%,85.98% and 93.24% respectively.(3)In recent years,music therapy has been widely used.The aim of this paper is to analyzes the effect of music therapy on the fetal heart rate through music therapy experiments for pregnant women.First of all,118 pregnant women of 32 to 40 weeks of gestational age were invited to participate in the experiment.We have a normal group and three experimental groups.The four groups were the normal group,the experimental group who listened to music in the first ten minutes and did not listen to music in the last ten minutes,the experimental group who did not listen to music in the first ten minutes and listened to music in the last ten minutes and the experimental group who listened to music for twenty minutes.We records pre-experimental fetal heart rate characteristics of pregnant volunteers,including acceleration,deceleration,fetal heart baseline,amplitude variation,cycle variation,fetal movement and contractions,and then record the fetal heart rate characteristics after music treatment.After the difference analysis of paired sample T test,the acceleration,baseline,amplitude variation,cycle variation,fetal movement and contraction were significantly different(P<0.05,except for deceleration)after music therapy,and there was a certain improvement.At the same time,we use the K-means clustering algorithm to analyze the difference,and found that there is a significant difference in the baseline,baseline variation and acceleration(P <0.05).In conclusion,music therapy can significantly increase characteristic data of fetal heart rate.We developed the electronic fetal monitoring system in cooperation with the hospital and commercialized it.Through the system,we collect a great number of fetal heart rate data.And then we improve the accuracy of fetal heart rate classification by CNN.At the same time,we also study the impact of music on the FHR.By analyzing the experimental data,we found that music therapy can significantly improve the characteristic data of fetal heart rate,such as baseline,baseline variability and acceleration et al.