Personalized Anatomic Modeling for Noninvasive Fetal ECG: Methodology and Applications

Fetal cardiac monitoring is one of the cornerstones of modern obstetric care. Noninvasive fetal electrocardiography (NI-FECG) is an emerging modality for monitoring fetal well-being using electrical signals recorded from the maternal abdomen. However, the reliability of NI-FECG extraction techniques remains highly variable due to a range of technical and clinical factors, such as sensor placement and interindividual anatomic variations. In this work, we propose, develop, and validate an open-source method for modeling these variations, including changes in maternal body structure and fetal position using two clinical NI-FECG databases. To validate our model’s accuracy, we first assess its performance in characterizing the fetal QRS (fQRS) complex amplitude in six private NI-FECG recordings with detailed anatomic information. To demonstrate its clinical utility, we next apply our model to predict an optimal sensor placement in a separate open-access database of 60 24-channel NI-FECG recordings. The optimal six sensor positions predicted by our model achieve similar reliability for fetal heart rate (FHR) monitoring compared to the entire 24-sensor array. The presented results indicate our model provides a suitable method for estimating the influence of anatomic variations on NI-FECG signals and optimizing sensor placement in a simulated setting. The code for the developed model has been made available under an open-source GPL license and contributed to thefecgsyntoolbox.