Unveiling fetal heart health: harnessing auto-metric graph neural networks and Hazelnut tree search for ECG-based arrhythmia detection

• Published in: Computer methods in biomechanics and biomedical engineering Vol. 28; no. 10; pp. 1671 - 1684
• Main Authors: Suganthy, M., Sarala, B., Sumathy, G., Chembian, W. T.
• Format: Journal Article
• Language: English
• Published: England Taylor & Francis 27.07.2025
• Subjects: Algorithms; Arrhythmias, Cardiac - diagnosis; Arrhythmias, Cardiac - physiopathology; Auto-metric graph neural network; dual tree complex wavelet transform; ECG signals; Electrocardiography; Female; fetal arrhythmia detection; fetal ECG data base; Fetal Heart - physiopathology; Graph Neural Networks; Humans; Neural Networks, Computer; Pregnancy; Signal Processing, Computer-Assisted; Wavelet Analysis; dual tree complex wavelet transform; ECG signals; fetal arrhythmia detection; fetal ECG data base; Auto-metric graph neural network
• ISSN: 1025-5842; 1476-8259; 1476-8259
• DOI: 10.1080/10255842.2025.2481232

Fetal electrocardiogram (ECG) provides a non-invasive means to assess fetal heart health, but isolating the fetal signal from the dominant maternal ECG remains challenging. This study introduces the FHH-AMGNN-HTSOA-ECG-AD method for enhanced fetal arrhythmia detection. It employs Dual Tree Complex Wavelet Transform for denoising and utilizes an Auto-Metric Graph Neural Network (AMGNN) optimized by the Hazelnut Tree Search Algorithm (HTSOA). This integration enables accurate classification of normal and abnormal fetal heart signals. Experimental results demonstrate that the proposed approach significantly outperforms existing methods in terms of accuracy, precision, and specificity.