Contactless Fine-grained Cardiac Events Detection and Segmentation with Radio Frequency Signals
Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies · 03.09.2025 · 9 (3) · 1-28
DOI: 10.1145/3749457 ↗Kurz gefasst
Fachartikel in der Zeitschrift Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, online erschienen am 03.09.2025. Verfasst von Zehan Guo und 4 weiteren Autor:innen. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 84 Literaturangaben.
Abstract (Originalsprache)
The rising prevalence of cardiovascular diseases has created an urgent need for accurate, long-term cardiac monitoring. Traditional electrode-based monitoring devices, which require direct skin contact, can cause discomfort and inconvenience. Although RF-based contactless sensing offers a promising alternative, current solutions struggle to accurately detect and segment fine-grained cardiac events, which are crucial biomarkers for cardiac function assessment and disease diagnosis. Due to spectral leakage caused by respiratory motion and the difficulty in capturing subtle mechanical motions within a single heartbeat cycle, detection and segmentation often suffer from temporal boundary ambiguity and critical event omission. To address these issues, we propose an innovative millimeter-wave (mmWave) fine-grained cardiac event detection and segmentation system that integrates a multi-scale boundary enhancement framework. This framework consists of a multi-scale context fusion module and a boundary-aware enhancement module, which collaboratively mitigate respiratory motion interference and improve feature recognition capabilities. Through dedicated training, our system captures both multi-scale local details and global relationships, enabling user-independent detection and segmentation without the need for subject-specific calibration. The experiments in outpatient of a hospital involving 4,339 participants, including participants with conditions such as Bundle Branch Block (BBB), First-Degree Atrioventricular Block (I-AVB), and T wave abnormalities (T abn.) show that the proposed system achieves an F1 score of 90.6%, outperforming baseline methods, and simultaneously achieves an average estimation error of 14.9 milliseconds for cardiac event intervals, representing a reduction of 5.1 milliseconds compared to the baseline error. The results demonstrate that this system can robustly detect fine-grained cardiac events, thereby providing a more reliable technical foundation for early cardiovascular monitoring.
Zitieren
APA
Guo, Z., Zhang, B., Chen, J., Hu, Y., & Chen, Y. (2025). Contactless Fine-grained Cardiac Events Detection and Segmentation with Radio Frequency Signals. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, 9(3), 1-28. https://doi.org/10.1145/3749457
Vancouver
Guo Z, Zhang B, Chen J, Hu Y, Chen Y. Contactless Fine-grained Cardiac Events Detection and Segmentation with Radio Frequency Signals. Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies. 2025;9(3):1-28. doi:10.1145/3749457
BibTeX
@article{Guo2025,
author = {Guo, Zehan and Zhang, Binbin and Chen, Jinbo and Hu, Yang and Chen, Yan},
title = {Contactless Fine-grained Cardiac Events Detection and Segmentation with Radio Frequency Signals},
journal = {Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies},
year = {2025},
volume = {9},
number = {3},
pages = {1--28},
doi = {10.1145/3749457}
}
EUVIMED