Fetal Arrhythmia Detection Based on Labeling Considering Heartbeat Interval.

Area of Science:

• Biomedical Engineering
• Cardiology
• Artificial Intelligence

Background:

• Fetal arrhythmia is a critical factor in sudden infant death, necessitating accurate detection for fetal well-being.
• Fetal electrocardiogram (FECG) analysis is crucial for detecting heart rhythm abnormalities, but its accuracy is often limited by heartbeat detection precision.

Purpose of the Study:

• To develop and validate a deep learning-based method for accurate fetal arrhythmia detection using FECG signals.
• To overcome the limitations of conventional methods that rely heavily on precise heartbeat detection.

Main Methods:

• The proposed method segments FECG signals and utilizes a deep learning model for classifying segments as normal or arrhythmic.
• Training data segments are categorized based on heartbeat intervals, with only clearly normal or arrhythmic segments used for model training to enhance classification accuracy.
• The method processes multiple segment classifications to determine the overall health status of the fetus.

Main Results:

• The deep learning model achieved 96.2% accuracy in detecting fetal arrhythmia.
• The method demonstrated 100% specificity and 100% recall, indicating superior performance.
• The proposed approach showed significant improvements over conventional methods reliant on heartbeat and feature detection.

Conclusions:

• Deep learning offers a robust solution for fetal arrhythmia detection from FECG signals, mitigating reliance on precise heartbeat detection.
• The developed method provides highly accurate and reliable detection of fetal arrhythmia, contributing to improved fetal health monitoring and potentially reducing risks associated with sudden infant death.