Fetal heart rate baseline computation with a weighted median filter

Area of Science:

• Biomedical Signal Processing
• Maternal-Fetal Medicine
• Computational Physiology

Background:

• Automated fetal heart rate (FHR) analysis aims to reduce inter- and intra-expert variability in interpretation.
• Accurate FHR analysis is crucial for identifying fetal acidosis and preventing unnecessary medical interventions.
• Key steps include baseline determination and detection of accelerations and decelerations (A/D).

Purpose of the Study:

• To introduce and evaluate a novel weighted median filter baseline (WMFB) method for automated FHR analysis.
• To compare the WMFB method's performance against existing literature methods and expert consensus.
• To assess the agreement between the WMFB method and expert interpretations of FHR recordings.

Main Methods:

• A weighted median filter baseline (WMFB) was computed for FHR signals.
• Filter weightings were determined by the prior probability of the FHR being in a baseline or A/D state.
• Performance was evaluated using a dataset of 90 annotated FHR recordings, comparing against expert consensus and 11 other methods using the morphological analysis discordance index (MADI).

Main Results:

• The WMFB method demonstrated significantly lower disagreement (4.02% MADI) compared to the best literature method (7.27%).
• The WMFB method's agreement with expert consensus was not significantly different from inter-expert agreement (p=0.22).
• The WMFB method outperformed 11 other published methods in reproducing expert consensus FHR analysis.

Conclusions:

• The WMFB method provides an automated FHR analysis that accurately reflects expert consensus.
• The method's performance is comparable to individual expert interpretation, offering a reliable tool for clinical use.
• The WMFB method shows significant promise for improving the objectivity and accuracy of FHR interpretation.