Multivariate Time–Frequency Analysis of Electrohysterogram for Classification of Term and Preterm Labor

Abstract

Non-invasive electrohysterogram (EHG) could be a promising technique for the preterm birth prediction, which could enable us to diagnose the preterm birth before the labor and reduces the infant mortality and morbidity. Previous studies on the preterm birth prediction with EHG have conducted comprehensive researches on various signal features and classification algorithms, but most of them adopted prefilters based on the linear transforms using fixed basis function, although they are suboptimal for the nonlinearity and nonstationarity of the EHG signal. In this paper, multivariate empirical mode decomposition (MEMD) is applied to decompose the electrical activity signal measured on the uterus. After the decomposition, features are calculated for the corresponding oscillations to the uterine contraction. To investigate the performance of the features, three-channel EHG signals of 254 patients (224 term, 30 preterm) are chosen among 300 patients from Physionet term-preterm electrohysterogram (TPEHG) database to extract features from the EHG signals and classify the features using machine learning algorithms. Classification results shows that the proposed method with MEMD achieved 94.66% correctly classified rate (CCR) and 0.987 area under the curve (AUC), which outperformed those with IIR filter implying MEMD provides a new prospect to improve the current preterm birth prediction approach.

Electrohysterogram for ANN-Based Prediction of Imminent Labor in Women with Threatened Preterm Labor Undergoing Tocolytic Therapy

“…Moreover, most of the present EHG recordings (≈84%) were taken from patients either undergoing or who had received tocolytic treatment. In this respect, tocolytic drugs such as Atosiban, which is a calcium channel blocker commonly used to inhibit uterine contractions, has been proven to shift signal spectral content to lower frequencies-especially in women closer to delivery-thereby reducing PSD peak frequency and/or the high-to-low frequency content ratio [55]. We consider that these factors could be the main reason for the disagreement between our present results with others in the literature.…”

“…The AUC for predicting labor in less than 7 days was of 87.1 ± 4.3% (test dataset), which can be considered a more realistic expected performance for predicting imminent labor for the new incoming data. The result achieved in the present work is slightly inferior to the model performance for testing data for predicting preterm labor in women in regular checkups by EHG (AUC~91%) [33], which can be attributed to the influence of the tocolytic drug on the EHG signal [26,55].…”

Optimization of Imminent Labor Prediction Systems in Women with Threatened Preterm Labor Based on Electrohysterography

“…ELM F1_2 metrics are slightly higher than those previously obtained using the same EHG recording database with an ANN classifier, for both validation and test groups (F1-score of 84.3 ± 5.0%, sensitivity of 86.5 ± 7.4% and specificity of 81.5 ± 7.3 for the validation dataset with ANN and F1-score of 80.3 ± 5.5%, sensitivity of 81.6 ± 9.4% and specificity of 78.8 ± 5.8% for the test dataset) [ 27 ]. Indeed, AUC values for ELM F1_2 in validation and test were over 93%, similar to those reported in the literature for preterm labor predictive systems based on EHG recordings during regular checkups [ 12 , 21 , 22 , 31 , 37 , 55 , 58 ], and slightly higher than those achieved in imminent labor prediction in women with TPL using ANN (AUC validation 91.8 ± 3.2%, AUC test 87.1 ± 4.3%) [ 27 ]. This could be due to the fact that in order to avoid overfitting, the two optimization criteria were applied to the validation dataset, whereas in our previous work the square root of the training and validation F1-score was optimized [ 27 ].…”

Assessment of Features between Multichannel Electrohysterogram for Differentiation of Labors

“…Many features have been extracted from EHG signals to recognize preterm birth. Time, frequency, and time-frequency features [14] such as root mean square, median frequency, peak frequency, energy distribution, etc., have been used to characterize EHG signals. Previous studies found that uterine activities are nonlinear processes that change with time, and nonlinear signal processing techniques could thus provide additional information on physiological changes during pregnancy and close to delivery.…”