Enhancing classification of preterm-term birth using continuous wavelet transform and entropy-based methods of electrohysterogram signals

Abstract

IntroductionDespite vast research, premature birth's electrophysiological mechanisms are not fully understood. Prediction of preterm birth contributes to child survival by providing timely and skilled care to both mother and child. Electrohysterography is an affordable, noninvasive technique that has been highly sensitive in diagnosing preterm labor. This study aimed to choose the more appropriate combination of characteristics, such as electrode channel and bandwidth, as well as those linear, time-frequency, and nonlinear features of the electrohysterogram (EHG) for predicting preterm birth using classifiers.MethodsWe analyzed two open-access datasets of 30 minutes of EHG obtained in regular checkups of women around 31 weeks of pregnancy who experienced premature labor (P) and term labor (T). The current approach filtered the raw EHGs in three relevant frequency subbands (0.3–1 Hz, 1–2 Hz, and 2–3Hz). The EHG time series were then segmented to create 120-second windows, from which individual characteristics were calculated. The linear, time-frequency, and nonlinear indices of EHG of each combination (channel-filter) were fed to different classifiers using feature selection techniques. ResultsThe best performance, i.e., 88.52% accuracy, 83.83% sensitivity, and 93.22% specificity, was obtained in the 2–3 Hz bands using Medium Frequency, Continuous Wavelet Transform (CWT), and entropy-based indices. Interestingly, CWT features were significantly different in all filter-channel combinations. The proposed study uses small samples of EHG signals to diagnose preterm birth accurately, showing their potential application in the clinical environment.DiscussionOur results suggest that CWT and novel entropy-based features of EHG could be suitable descriptors for analyzing and understanding the complex nature of preterm labor mechanisms.

Automated Method for Uterine Contraction Extraction and Classification of Term Versus Pre-Term EHG Signals

“…Therefore, in addition to these classical parameters related to contractile activity, entropy indices were extracted from contraction segments to assess the efficacy of the method in classifying between normal and premature births. Recently, several academics have concentrated on entropy-based indices as a means of evaluating and quantifying the complexity of EHG time series [57][58][59][60][61]. Most of the researchers suggested that for predicting preterm labor and assessing the progress of labor efficiently it would be better to choose sample entropy and median frequency as potential features.…”

An open dataset with electrohysterogram records of pregnancies ending in induced and cesarean section delivery

“…In 2021, an important study 26 revealed that over-sampling applied after data partitioning, i.e., partition-synthesis over-sampling approach, needs to be applied to achieve realistic classification performance, and realistic preterm birth prediction in the case of imbalanced sets. Recently, many interesting studies related to preterm birth prediction using the TPEHG DB were published using traditional feature engineering 27 – 33 and deep learning 34 – 37 approaches. A nice review of the literature dealing with the use of EHG records for the task of predicting premature birth and for understanding the underlying physiological processes during pregnancy can be found in 38 .…”

“…Merging EHG records from different databases/dataset may be questionable due to the differences in signal acquisition protocols. Since the EHG records of the TPEGH DB and TPEHGT DS were acquired under the same acquisition protocol, and using the same recording device, several authors merged the EHG records from these two database/dataset in cases of traditional feature engineering 24 , 25 , 32 , 33 and deep learning approach 37 .…”

Prediction of Preterm Labor from the Electrohysterogram Signals Based on Different Gestational Weeks

“…Second, reducing the number of employed channels, e.g., a single EHG channel, should be investigated as welcome in longterm home-based pregnancy monitoring systems. Third, the effectiveness of the proposed method should be further investigated using different versions of filtered EHG signals (e.g., 0.3 to 3 Hz) or even with different frequency ranges as suggested in [53]. Fourth, though the selected features showed acceptable performance, it should be noted nonetheless that the employed strategy might not exclude the redundant features.…”