Contraction extraction from term and preterm electrohyterographic signals

Rates of mortality and morbidity of premature newborns would be considerably reduced if threats of preterm labor could be predicted few hours or even few days before. The aim of this study is the analysis of non-invasive ElectroHysteroGraphic (EHG) records for prediction of preterm deliveries. We proposed a procedure to automatically extract pregnancy contractions from long duration EHG recordings because the study of EHG contractions is more efficient in discriminating between term and preterm ending pregnancies. The proposed method is based on variations of Kalman filter innovation of an autoregressive moving average (ARMA) model for EHG record. The procedure was applied on real EHG data. The term and preterm characteristics of time and frequency related features resulting from the extracted contractions were in accord with those in literature.