Related Experiment Video
Updated: Jan 15, 2026
07:14
Tracking Infiltration Front Depth Using Time-lapse Multi-offset Gathers Collected with Array Antenna Ground Penetrating Radar
Published on: May 1, 2018
8.2K
Estimation of Compression Depth During CPR Using FMCW Radar with Deep Convolutional Neural Network.
Insoo Choi1, Stephen Gyung Won Lee2, Hyoun-Joong Kong2
1Department of Electronic Engineering, Sogang University, Seoul 04107, Republic of Korea.
Sensors (Basel, Switzerland)
|October 16, 2025
Summary
This study introduces frequency-modulated continuous-wave (FMCW) radar for accurate, remote chest compression depth monitoring during cardiopulmonary resuscitation (CPR). A Wigner-Ville distribution-based deep learning model achieved the highest accuracy, improving emergency medical response.
Area of Science:
- Biomedical Engineering
- Signal Processing
- Artificial Intelligence
Background:
- Effective cardiopulmonary resuscitation (CPR) demands precise chest compression depth.
- Current out-of-hospital monitoring technologies for CPR lack accuracy and have limitations.
- Accurate, real-time measurement of chest compression depth is crucial for improving CPR outcomes.
Purpose of the Study:
- To develop and evaluate a non-contact method for accurately measuring chest compression depth during CPR using FMCW radar.
- To compare the performance of different signal processing and deep learning techniques for analyzing radar data.
- To demonstrate the potential of FMCW radar and AI in enhancing emergency medical response.
Main Methods:
- Utilized frequency-modulated continuous-wave (FMCW) radar to capture range, Doppler, and angular data of chest movements.
- Employed micro-Doppler signatures and integrated Doppler shifts to estimate chest displacement.
- Compared a regression model with deep convolutional neural networks (DCNNs) trained on spectrograms from STFT and WVD.
Main Results:
- The regression model achieved a root mean square error (RMSE) of 0.535 cm.
- The STFT-based DCNN improved accuracy with an RMSE of 0.505 cm.
- The WVD-based DCNN achieved the best performance with an RMSE of 0.447 cm, an 11.5% improvement over the STFT-based DCNN.
Conclusions:
- FMCW radar combined with deep learning, particularly using WVD, offers a promising approach for accurate, non-contact chest compression depth measurement.
- This technology has the potential to significantly improve CPR quality and patient outcomes in out-of-hospital settings.
- The study supports the development of advanced, real-time monitoring systems for emergency medical services.
