Digital Method Of Binary CTG Paper Report Based On Smart Phone

Cardiotocography(CTG),including continuous monitoring of fetal heart rate(FHR;bpm)signals and maternal uterine contractions(UC;mm Hg)signals,is the most effective method for early detection and treatment of fetal cerebral palsy and preterm birth caused by intrauterine ischemia and hypoxia.Clinically,CTG reports are usually output on paper and visually interpreted by clinicians to assess fetal health.However,the paper CTG report is very restricted in storage,transmission,retrieval and sharing among doctors due to its own characteristics.More importantly,due to the lack of public and open CTG data resources,this greatly hinders the development of intelligent CTG signals analysis systems based on deep learning.Therefore,digitizing CTG paper reports can not only compensate for the inconvenience of storing and sharing CTG paper reports,but also enrich the CTG database.The existing CTG paper report digitization method extracts signals from colored background grid lines,but it cannot be extracted from the binary CTG paper report commonly used in clinical practice.Moreover,the existing method uses a scanner to obtain the digital image of the CTG paper report,which requires a computer to complete the digitization and intelligent analysis of the CTG paper report,which is poor in portability.In order to overcome the shortcomings of existing methods and the convenience of using smart phones to obtain digital images,this thesis proposes a digital method of binary CTG paper reports based on smart phones.Firstly,smart phones are used to obtain digital images of binary CTG paper reports and preprocess the digital images.Then,the FHR signals and UC signals are regionally located.Next,the background grid lines of the binary CTG paper report were removed and the signal lines were reconstructed.Finally,the error caused by the distortion of the mobile phone lens is distorted.The main research content of the thesis is as follows.1.An adaptive area location algorithm based on statistical calculations is designed to locate FHR and UC signal areas.2.Design a method based on statistical calculation,connected domains and weight sum to remove the background grid lines of the CTG report.3.A calibration method based on segmented sampling is proposed to reduce the error caused by the distortion of the smartphone lens.The digital signal obtained by the method in this thesis is compared with the reference signal.The root mean square error(RMSE),correlation coefficient(ρ)and mean absolute error(MAE)of the FHR signal are 1.18 bpm,0.98 and 0.87 bpm,respectively.The root mean square error,correlation coefficient and mean absolute error of the UC signal are 2.10 mm Hg,0.98 and 1.50 mm Hg respectively.