QRS detector for maternal abdominal ECG

Abstract

Detection of QRS complexes of the Electrocardiogram (ECG) signals plays a significant role in detecting heart rates and diagnosing most of the cardiac diseases. The abdominal ECG (aECG) recorded from the maternal abdominal surface of a pregnant woman is a composite of Maternal ECG (MECG), Fetal ECG (FECG), Electromyogram (EMG), power line interference, baseline drift, motion artifact and electrode contact noise. Monitoring of maternal and fetal ECG of pregnant mothers from the collected aECGs can detect accurate MECG and can also detect fetal asphyxia early in the evolution to acidosis. A modified QRS detection algorithm based on the Pan Tompkins algorithm was optimized to minimize the number of false QRS detections. The algorithm performance is evaluated with the physionet databases 1) Abdominal and direct fetal ECG (adfecgdb) and 2) Physionet Challenge 2013 (PhyC). The algorithm adjusts the threshold for each record to obtain the Maternal Heart Rates (MHR) giving an error of 0.012 percent and 0.044 percent respectively for the two databases. This algorithm correctly detects Sensitivity (Se) and Positive Predictivity (PP) of 99.49 percent and 99.35 percent respectively for PhyC records while, 100 percent of Sensitivity and Positive Predictivity for the adfecgdb.