Fetal Heart Rate Extraction from Abdominal Electrocardiography Recordings Based on Wavelet Transform and Adaptive Threshold Algorithm

Fetal heart rate monitoring during pregnancy can help to diagnose distress and morbidity for a fetus. Noninvasive fetal electrocardiography (NI-FECG) is a promising technology that provides critical information about a fetus's health status. Our study applies the template subtraction (TS) algorithm for fetal ECG signal extraction from abdominal ECG (AECG) recordings. The combination of discrete wavelet transform (DWT) and adaptive double threshold (ADT) algorithm is utilized to detect QRS waves for fetal ECG signals. To verify the algorithm's effectiveness, the Cardiology Challenge 2013 database (PCDB) and abdominal and direct fetal ECG database (ADFECGDB) are employed. The proposed method detects 8999 fetal QRS complexes, wrongly locates 707 fetal QRS complexes, and miss detects 583 fetal QRS complexes for the PCDB. The diagnostic Se, PPV, and F1 scores are 93.92%, 92.72%, and 93.31% of the PCDB. In addition, our work detects 3191 fetal QRS complexes, wrongly locates 62 fetal QRS complexes, and miss detects 38 fetal QRS complexes for AECG recordings of the ADFECGDB. Diagnostic Se, PPV, and F1 scores are 98.81%, 98.07%, and 98.44% of the ADFECGDB. The results show that our proposed method can improve the accuracy of fetal QRS complex detection to a certain extent.