Evaluation of an improved algorithm for fetal QRS detection

In the area of fetal heart rate (FHR) monitoring, different sizes and multifunctional ultrasound based systems are used. All these systems are invasive but cannot be use for long term monitoring applications. The objective of this paper is to evaluate a fetal peak detection algorithm which is applied to all the abdominal channels without threshold (amplitude threshold free) that is threshold independent detection. The proposed algorithm detects fetal R peaks from pregnant abdominal noninvasive records. Fetal signals were extracted using recursive least square (RLS) adaptive filter. Subsequently an enhancement technique based on maternal QRS (MQRS) window was applied to the residual signal, then the signal smoothed by attenuating other component around one Hz using IIR notch filter. MQRS peaks were adjusted in order to make it shorter than fetal QRS (FQRS) peaks. Therefore, peak position correction is also performed to correct false detections (overlapped peaks) using correction technique based on the window signal. Thirty recorded signals (each of 60 seconds) were acquired from pregnant women (from 36 to 38 weeks of gestation) in order to evaluate the proposed algorithm. A sensitivity of 79.8% and a positive prediction value of 77.5% were obtained. The error rate of the proposed algorithm is in the range of 0.7 to 4.9% compare with the ultrasound measurements.