Automated detection of perinatal hypoxia using time-frequency-based heart rate variability features

Area of Science:

• Biomedical Engineering
• Neonatal Medicine
• Signal Processing

Background:

• Perinatal hypoxia can cause significant cerebral injury in fetuses and neonates.
• Current methods for detecting fetal hypoxia during labor, relying on heart rate signal pattern recognition, exhibit high observer variability and low specificity.
• There is a critical need for more accurate and reliable methods for intrapartum fetal hypoxia detection.

Purpose of the Study:

• To develop and validate a novel automated method for detecting fetal hypoxia using time-frequency analysis of heart rate variability (HRV) signals.
• To improve the accuracy and reduce observer variability in fetal hypoxia detection during labor.
• To establish a more reliable tool for human fetal monitoring.

Main Methods:

• Utilized time-frequency analysis of heart rate variability (HRV) signals.
• Extracted features from instantaneous frequency and amplitude of HRV signal components.
• Employed matrix decomposition techniques (SVD, NMF) and a support vector machine classifier for hypoxia detection.
• Validated the method on a newborn piglet model with controlled hypoxic insult.

Main Results:

• The proposed HRV feature-based method demonstrated high performance in hypoxia detection.
• Achieved a sensitivity of 93.3%, specificity of 98.3%, and accuracy of 95.8% in the piglet model.
• Outperformed conventional spectral features and existing hypoxia detection methods.

Conclusions:

• The developed automated hypoxia detection method shows substantial promise for clinical application.
• This approach offers a significant advancement towards more accurate and reliable intrapartum fetal monitoring.
• The high predictive value supports its potential to improve neonatal outcomes by enabling timely intervention.