Fetal QRS complex detection based on adaptive filters and peak detection

• Published in: Research on biomedical engineering Vol. 41; no. 3; p. 42
• Main Authors: Rad, Elias Mazrooei, Zendehbad, Seyyed Ali, Hosseinzadeh, Vahideh
• Format: Journal Article
• Language: English
• Published: Cham Springer International Publishing 01.09.2025; Springer Nature B.V
• Subjects: Accuracy; Adaptive filters; Algorithms; Arrhythmia; Artificial intelligence; Biomaterials; Biomedical Engineering and Bioengineering; Biomedical Engineering/Biotechnology; Butterworth filters; Cardiac arrhythmia; Complexity; Computer applications; EKG; Electrocardiography; Engineering; Epidemiology; Error detection; Fetuses; Filters; Gestational age; Heart; Heart rate; Identification; Infant mortality; Innovations; Least mean squares; Least mean squares algorithm; Methods; Morphology; Noise reduction; Original Article; P waves; Real time; Robustness; Signal processing; Signal quality; Waveform analysis; Waveforms; Peak thresholds; Adaptive filters; Fetal electrocardiography (FECG); QRS complex detection; ECG monitoring
• ISSN: 2446-4732; 2446-4740
• DOI: 10.1007/s42600-025-00424-4

Purpose The proposed research presents a new algorithm for Non-Invasive Fetal Electrocardiography (NI-FECG) which evaluates all ECG signal components starting from the P wave through the PR interval and QRS complex to ST segment and T wave and U wave and QT interval. Correctly identifying these components is vital for achieving a complete assessment of fetal heart health. A complete diagnostic evaluation needs the analysis of all waveform components even though detecting arrhythmias and guiding clinical interventions depends primarily on the QRS complex. Least mean squares (LMS) adaptive filters along with Butterworth filters operate within the algorithm to automatically reduce noise while improving signal quality using adaptive filtering. The detection system uses dynamic peak thresholding for adjusting thresholds according to signal characteristics so it provides stable QRS complex identification in noisy conditions. The proposed algorithm demonstrates strong computational efficiency and robust performance which makes it ideal for real-time clinical implementation and separates it from other typical methods. Materials and methods The proposed approach analyzes 55 NI-FECG multichannel recordings obtained from pregnant women between weeks 21 and 40. Dynamic peak thresholding with adaptive filtering along with detection accuracy enhancement techniques forms the basis of this method to reduce noise disturbances and improve ECG signal clarity during different components of detection. The research examines ECG waveform variations together with arrhythmic effects on QRS morphology in great detail. Results Experimental findings demonstrate that the proposed algorithm maintains robustness through a detection error average of 1.78% accompanied by a standard deviation of 0.46% which was tested across all participants when evaluating all ECG components, particularly QRS detection. Real-time clinical deployments can be supported by the proposed algorithm thanks to its reduced computational complexity as well as improved detection accuracy. The algorithm operates with high precision while using only 0.22 s to execute which makes it suitable for real-time application. Conclusion Research has developed an outstanding algorithm to analyze NI-FECG signals which delivers exceptional precision for detecting all morphological attributes including PR, RR, ST, and QT intervals. The holistic examination technique enables better accuracy in fetal heart health assessments. The analysis of fetal ECG signals exhibits stronger accuracy and robustness to noise when biological processing methods are specifically developed to replace generic artificial intelligence models. The development of clinical performance optimization requires additional research across different medical circumstances. The implementation of this method in regular clinical care would boost fetal heart health evaluation and produce better pregnancy results.