Fractional order calculus for fetal pulse detection

Abstract

This paper presents an extension and modification of the Pan-Tompkins (PT) algorithm for fetal QRS detection with the application of fractional order calculus. The novel method was applied for fetal QRS detection in electrocardiogram (ECG) recorded noninvasively from the abdomen of pregnant woman - in an ECG signal available from an open database. Algorithm robustness was tested on ECG signals with additive Gaussian noise with 8 different Signal-to-Noise Ratios (SNR). The optimal fractional order allows relatively accurate detection of fetal QRS complexes for different SNRs. The performance of the algorithm was evaluated by three parameters: root mean square error, percentage of missed QRS complexes, and the amplitude ratio of fetal and pregnant woman's ECG. The presented algorithm can be used for assessment of fetal heart rate variability which is important for timely detection of potential fetal heart disorders as well as in similar spike-shape signals in the presence of noise components.