Research On Methods Of Cardiotocographic Parameter Analysis And Fetal State Assessment

With the constant rise of the rate of birth defect in China in recent years, the one-child policy and the prenatal and postnatal care policy have faced severe challenges. In order to reduce the rate of birth defect and infant mortality, fetal monitoring is carried out to assist medical staffs in detecting the potential fetal healthy problems. Cardiotocography(CTG) is a common method for fetal monitoring, by which CTG signals are obtained. Medical staffs can assess the fetal state by analyzing CTG signals. However, as a matter of fact, many medical staffs lack enough capabilities of analysis and judgment of the CTG signals and consequently make inappropriate clinical decisions easily. The appearance of CTG computer-aided analysis can alleviate the problem to some extent. Nonetheless, due to the complexity of CTG signal and the impreciseness of the definition of CTG parameters, computerized analysis still has numerous problems, such as the poor accuracy of automated parameter identification. In order to improve the accuracy and efficiency of parameter identification, the researches described in this thesis were devoted to the methods for extracting some key parameters from CTG signals. Based on these, the method for fetal state assessment was studied and consequently a more effective evaluation method was proposed.The main contents of this thesis are as follows:(1) The research on fetal heart rate(FHR) baseline estimation algorithm. In this thesis, an FHR baseline estimation algorithm combined with fetal movement was presented to solve the problems on the accuracy or efficiency of the existing algorithms. In the proposed method, the location of acceleration in the FHR tracing is determined by fetal movement before baseline estimation. Then baseline is estimated and corrected after the accelerations are erased. To evaluate the performance of the proposed baseline estimation algorithm, a contrast test was designed to compare it with two existing algorithms. The results showed that: in the respect of analysis accuracy, the overall performance of the proposed algorithm was significantly better than that of control algorithm I, and a little better than that of control algorithm II; and in terms of computational efficiency, the proposed algorithm is no different than control algorithm I, but far exceed control algorithm II.(2) The study on feature extraction algorithms for tocography(TOCO) tracing. This thesis studied the uterine contraction(UC) baseline estimation and UC wave identification and offered some new approaches. Compared with a previous method, the new UC baseline estimation algorithm can acquire a smoother baseline and achieve the breakpoint detection to obtain a reasonable final result. On the basis of the principle of image dilation and the morphology analysis of UC wave, this thesis proposed a UC identification algorithm. In comparison to a previous UC identification algorithm, it was observed that the performance of the proposed algorithm was better than the control algorithm, showing as a lower false positive rate and false negative rate. Meanwhile, a method for the UC status real-time identification was implemented, which can achieve real-time recognition of four kinds of UC status. Compared the real-time analysis results of the proposed algorithm with the doctor’s post-hoc analysis results, it was clear that the conformity between these two results was high. Besides the proposed method meets the requirements of real-time analysis on computing efficiency.(3) The research on the method for fetal state assessment. Facing the present problems existing in fetal state computerized analysis, mainly in poor accuracy which is caused by directly applying CTG classification criteria in CTG computer-aided analysis system, this thesis properly adjusted the analysis criteria based on the literature research productions and experimental results. In the proposed method, fetal state assessment is achieved by calculating the CTG parameters’ membership degrees to three states referencing to the idea of fuzzy set and measuring the differences between CTG signal and three standard states which is indicated by Euclidean distance. To evaluate the effectiveness of the method, a control experiment against a computer analysis method which directly applied the CTG classification criteria was designed. The results showed that the proposed method can identify much more normal CTG signals. In addition, the proposed method had much higher specificity and positive predictive value than control method, and the overall accuracy was also greatly higher than that of control method.