A comprehensive adaptive filtering algorithm for uterine contraction pressure signals

Monitoring uterine contraction (UC) events provides a significant means to control the process of labor. Filtering UC signals is an indispensable step before UC identification through automated fetal monitor system (FMS) analysis. A comprehensive adaptive filtering algorithm based on signal quality determination and signal classification is proposed in this paper. After estimating the signal quality, the signal is classified into a status based on the determined quality. Meanwhile, a suitable filtering method is selected adaptively. The experiments show that the conformity between the final analysis results via the adaptive filtering method and experts' analysis are comparatively higher than when one constant filtering method is used. The adaptive way achieved sensitivity (Se) above 93% and positive predictive value (PPV) up to 94.08% for UC records in the testing set, while the average Se and PPV of the three single-filtering means are separately 78.92% and 86.51%. Other signal-filtering means used achieved lower values. Therefore, the proposed method can satisfy the requirements for automated MFS analysis with improvement in both efficiency and accuracy.