Does an on-device NEO-based adaptive predictor accurately detect fetal R-peaks and reduce data transmission in maternal-fetal heart monitoring from abdominal ECG?
A novel on-device signal processing architecture for fetal ECG monitoring achieves high accuracy and drastically reduces data transmission needs, enabling continuous wearable monitoring.
Existing non-invasive fetal ECG (FECG) systems often rely on external computation via wireless links, limiting their feasibility for long-term, battery-powered use. To overcome this limitation, a hardware-efficient signal processing architecture that performs fetal and maternal heart rate extraction fully on the device is presented. Key signal processing steps include parallel maternal/fetal band-pass filtering and a nonlinear energy operatorbased adaptive predictor to robustly identify maternal and fetal R-peaks in real time. Evaluated on public abdominal ECG datasets, the proposed on-device system achieved high fetal R-peak detection performance with an average F1-score of greater than 96.0%. Moreover, the system outputs processed results-specifically, fetal and maternal RR intervals-thereby reducing data transmission by greater than 99.9% compared to raw signal transmission. This fully on-device approach eliminates the need for high-data-rate wireless streaming, demonstrating its practical feasibility for continuous wearable FECG monitoring.
Cho et al. (Thu,) studied this question.