Comparative Measurement of Heart Rate and Respiratory Rate Using a Contactless Radar System

Magdalena Liebetruth^a,b, Philipp Stockel^c, Bassam Elmakhzangy^d, David Cyrol^a, Patrick Wallrath^c, Christophe Maufroy^d, Urs Schneider^d, Kai Kehe^e, Dirk Steinritz^f, Stefan Sammito^a,b,*

^a Department Research & Development, German Air Force Centre for Aerospace Medicine, Cologne, Germany

^b Institute of Occupational Medicine, Faculty of Medicine, Otto von Guericke University Magdeburg, Magdeburg, Germany

^c Fraunhofer Institute for High Frequency Physics and Radar Techniques (FHR), Fraunhoferstrasse 20, Wachtberg, Germany

^d Fraunhofer Institute for Manufacturing Engineering and Automation (IPA), Stuttgart, Germany

^e Bundeswehr Military Medical Academy, Munich, Germany

^f Bundeswehr Institute of Pharmacology and Toxicology, Munich, Germany

Summary

Background: Contactless radar technology for determining vital signs can be used in conjunction with drone technology to quickly locate and easily determine the vital signs of casualties in crisis and disaster situations. This experimental study investigated the accuracy and reliability of a contactless Multiple Input Multiple Output (MIMO)-Frequency-Modulated Continuous Wave (FMCW) radar system for measuring heart and respiratory rates, compared with an established contact-based reference system, in a static configuration as a first step toward future drone-based deployment.

Methods: Eleven subjects were measured under static conditions in five different body positions, with and without prior physical activity and with and without a military protective vest.

Results: The results showed a weak-to-moderate, statistically significant correlation (r =0.245, p < 0.001) in heart rate across all conditions and a weak, non-significant correlation in respiratory rate (r =0.169, p =0.063). In particular, wearing a protective vest had a negative influence on the correlation. Conclusions: Despite existing limitations, the radar system investigated here showed clear potential for use in realistic military medical and disaster medical scenarios.

Keywords: vital parameters; measurement; military medicine; disaster medicine; drone

Introduction

Measuring vital parameters such as heart rate and respiratory rate is crucial for assessing a person’s health. Normal values for adults are 60–100 beats per minute for the heartbeat and 12–20 breaths per minute for the respiratory rate. Traditionally, these parameters are measured using an ECG and breathing belt, but this is associated with restrictions on freedom of movement and patient comfort [10].

Contactless radar technology offers a promising alternative for non-invasive measurement of these parameters [10], especially in situations where it is not readily possible to monitor the patient with conventional vital-signs monitors. For example, this technology can be used to screen large numbers of injured people or to continuously monitor patients in hospitals. The combination of such radar technology for recording vital signs and a (small) drone is particularly suitable for use in disaster situations to quickly locate potential casualties and relay initial information about their health to rescuers on standby. Such a drone can also be used in a military setting to quickly identify possible casualties and thus provide qualified initial and follow-up care [7].

In a recently published systematic review [8], it was shown based on a total of 131 previously published papers that although acceptable results can be achieved when using radar technology to record vital signs, there are still challenges in terms of measurement accuracy, particularly in the case of involuntary body movements [9]. Furthermore, previous studies have only rarely investigated usability in a drone and/or combined this with military protective equipment for the participating test subjects, which could lead to possible limitations in the measurement accuracy of the radar technology due to the built-in Kevlar plates.

For this reason, a Multiple Input Multiple Output (MIMO)-­Frequency-Modulated Continuous Wave (FMCW) radar system was developed in the present experimental study to record heart rate and respiratory rate under different conditions. The aim was to evaluate the accuracy and reliability of the radar system in a static setting and under real conditions and to investigate its potential for use in various scenarios (including a military setting).

Materials and Methods

A total of eleven volunteers (nine male, two female, median age: 28 years [min: 20 years, max: 38 years]) took part in the present study. After a verbal briefing and explanation of the content and purpose of the study, a written declaration of consent to participate in the study and to data protection was obtained.

The test subjects were equipped with the mobPhysioLab physiological measurement system (KORA Industrie-Elektronik GmbH, Hambühren, Germany) prior to the measurements. In addition to using the so-called Heally Flash Master (HFM-01) to connect the measurement satellites and to store the data, the 3-channel ECG sensor (ECG-01) and a respiratory stretch sensor (airway sensor [AWS-01], respiratory band [E1227]) were used with the associated HealthLab satellites (SAT-02 and SAT-03).

The radar measurement was carried out in parallel with the contact measurement using an experimental system developed by Fraunhofer FHR, based on a commercial 76 GHz FMCW radar with a MIMO configuration of three transmitting and four receiving antennas [13]. It operates with a bandwidth of1 GHz to achieve a radar range resolution of15 cm to distinguish the human subject from the ground, while detecting millimeter- and sub-millimeter-scale movements through additional signal processing and a maximum measurement distance of 20 meters. It provided an azimuth field of view from -40° to +40° and an elevation field of view from -10° to +10°, with a measurement rate of 200 Hz. This setup facilitated precise detection of thoracic micro-movements corresponding to vital signs.

Respiratory and heart rates are estimated using a standard signal-processing chain, which would need to be extended with additional noise compensation and motion-­correction techniques when implemented on a drone in future work. The method is based on analyzing the phase of the radar signal reflected from a subject, where phase variations are induced by millimeter- and sub-millimeter-scale movements of the human body. The radar transmits a signal toward the subject and processes the reflected signal by extracting its phase. The signal is then passed through two band-pass filters tuned to the respiratory- and heart-rate frequency bands. Finally, a Fast Fourier Transform (FFT) is applied to each filtered signal, and the dominant frequencies are selected as the respiratory and heart rates.

As part of the data collection, 20 individual measurements were taken per test subject. The measurements were taken at 10 different positions, with each position measured twice to assess reproducibility. The duration of each measurement was one minute. The measurement protocol initially comprised basic positions without protective equipment: supine position (measurement 1–2), right-sided position (measurement 3–4), left-sided position (measurement 5–6), and prone position (measurement 7–8). After performing 20 jumping jacks, a further measurement was taken in the supine position (measurement 9–10). The complete series of measurements was then repeated using a military protective vest of the Bundeswehr type “Schutzweste Infanterie FTD 3” (measurements 11–20, same positions as without protective vest). The protective vest offers soft ballistic protection against handgun ammunition and shrapnel, as well as hard ballistic protection against large-caliber rifle ammunition. It is made of durable Cordura, has a modular design, and allows additional pockets and equipment to be attached using the MOLLE system or similar fastening options. The standard model weighs 13 kilograms. An overview is shown in Table 1.

Tab. 1. Overview of the different body positions during the measurement series with assignment to the measurements, once with and once without military protective vest

For all measurement positions, the test subjects lay at a radar-ground distance of 2.4 m below the radar, with the drone and the radar attached to it suspended from a stable structure (not flying). During each measurement, only one person was measured at the same time by the radar system (Figure 1). A one-minute measurement was performed at each position, and the data from the contactless radar system were compared with those from the mobPhysioLab. To ensure comparability of the measured values, the first 20 seconds of each measurement were discarded, as the radar system did not yet provide any readings during this initial phase. From the remaining 40 seconds, an arithmetic mean and the standard deviation of the measurement data were calculated for each subject and each position for each measurement system. Statistical analysis was done using IBM SPSS Statistics 24 (IBM, Armonk, NY, USA). Furthermore, a linear correlation analysis was carried out to compare the radar measurement with the reference measurement using the Pearson correlation coefficient, with r² specified to indicate the predictive power of the radar measurement for the reference measurement. The significance level was set at p < 0.05.

Figure 1. Picture of the test setting in the laboratory (up: schematically, down: real). (Copyright: Bundeswehr/ZentrLuRMedLw I 3)

 

Prior to the statistical comparative analysis, heart rates ≤ 40 and ≥ 130 and breathing rates ≤ 8 and ≥ 25, measured with the reference system, were excluded, as these values were considered unphysiological for the test setting. After this adjustment, 214 valid measurements remained for the heart rate (artifact rate: 6/220 = 2.7 %) and 122 for the respiratory rate (artifact rate: 98/220 = 44.5 %).

This study was conducted as part of the departmental research contract of the German Air Force Center for Aerospace Medicine in cooperation with the participating institutes. The ethics committee of the North Rhine Medical Association has given a positive vote on the conduct of the study (Ref. 2021438), and the study was positively evaluated by the data protection officer of the German Air Force Center for Aerospace Medicine.

Results

The comparison of heart rate measurements (n = 214) showed a weak-to-moderate, statistically significant correlation between the radar and reference systems (r = 0.245, p < 0.001). The radar measured higher values at 18 of 20 positions (90 %), except at positions 19 and 20, where the subject was examined after physical activity in the supine position wearing a military protective vest. On average, the radar measurement across all conditions was 81.8 ± 5.9 beats/min, while the reference measurement was 74.1 ±14.4 beats/min. In addition, the reference measurement had a higher standard deviation. The linear regression analysis showed that the radar system measurements could explain around 6 % of the variance in the reference measurements (R² = 0.060, F(1, 212) = 13.491, p < 0.001). The regression equation is: heart rate (mobPhysioLab) = 0.599 * heart rate (radar) + 25.152.

When differentiating the measured values under conditions without (position 01–10) vs. with protective vest (position 11–20), a correlation of r = 0.286 (p = 0.003) was observed for heart rate in the measurements without protective vest (n = 107). The mean values were 81.9 ± 5.6 beats/min for the radar measurement and 71.2 ± 12.7 beats/min for the reference system. With a protective vest (n = 107), a weaker correlation (r = 0.232; p = 0.016) was observed; the radar system measured 81.6 ± 6.2 beats/min, which was also higher than the reference system’s 77.0 ± 15.5 beats/min.

Looking at the measured values by measurement position, the highest correlations (r = 0.263 to 0.593) were observed in the supine position without the protective vest, while the lowest were observed in the prone position (r = -0.206 to 0.009). With a protective vest, the left-sided position showed the strongest correlation (r =0.290 and 0.542), whereas the lowest correlation was observed for the right-sided position (r = -0.261 and -0.018).

A graphical representation independent of the position is shown in Figure 2, and an overview of the heart rate measurements across all different positions is shown in Table 2.

The respiratory rate measurements (n = 122) showed a weak, non-significant correlation between the two measurement methods (r = 0.169, p = 0.063). Here, too, the radar system measured higher values on average (16.3 ± 4.8 breaths/min) than the reference system (14.3 ± 3.2 breaths/min). This was also observed in 90 % of the measurement positions. The exception here was also the measurement position after physical activity, but once with and once without a protective vest. In comparison, the radar measurement showed a higher standard deviation. The regression analysis of the respiratory rate yielded a very low R² of 0.029, indicating low predictive power of the radar measurements for the reference measurements (3 %). The regression equation here is breathing rate (mobPhysioLab) = 0.113 * breathing rate (radar) + 12.417. A significant correlation (r = 0.307, p = 0.016) was found in the respiratory rate measurements without a protective vest (n = 61), with an average of 17.1 ± 4.6 breaths/min for the radar system and 14.1 ± 3.4 breaths/min for the reference system. For measurements with a protective vest (n = 61), the correlation was not significant (r = 0.036, p = 0.786). Here, the radar system measured an average of 15.5 ± 4.8 breaths/min and the reference system 14.4 ± 3.0 breaths/min.

Looking at the measured values depending on the measurement position, the highest correlation values (r = 0.897 and 0.910) were again found in the supine position without the protective vest and the lowest for the prone position (r = -0.669 and -0.434). With a protective vest, the supine position was both the best correlated (r = 0.844, p = 0.004) and the least correlated (r = 0.082, p = 0.861).

A position-independent graphical representation is shown in Figure 3, and an overview of respiratory rate measurements across all positions is shown in Table 3.

 

Tab. 3. Comparison of the position-dependent measurements of the respiratory rate by radar and reference system

Discussion

The present study demonstrated the determination of vital signs using radar technology. Even though the accuracy of heart rate and respiratory rate measurements from the contactless radar system was lower than that of the established contact-based reference system, the system was still able to roughly detect vital signs, which is an additional value for use in disaster or military settings. The results for heart rate measurements were similar with and without wearing a military protective vest, whereas wearing the protective vest led to a significantly poorer correlation in respiratory rate measurements.

Correlation of heart rate

A weak but statistically significant correlation was found in heart rate across all conditions. The determined deviations are clinically significant in principle, but sufficient for a rough classification of vital signs. Compared with other FMCW radar studies, it is evident that the comparability of current studies is significantly limited by the highly heterogeneous approaches to statistical analysis [8]. The results of comparable studies are also heterogeneous, but there is generally good agreement among the measurement methods [11][12][14]. It is important to note that in the present study, compared to these studies, at least twice the distance of the subject to the radar was chosen, and in several of them, a lower radar frequency was used at the same time [14], both factors having a negative influence on the accuracy of the radar results and thus may explain a lower correlation [2][4].

With only 2.7 % of all measured data for the heart rate measurement that have been excluded in the present study indicates a reliable measurability of the reference system. The generally higher measured values in the radar measurement could indicate a systemic overestimation by the radar system, for example, due to the smallest, involuntary body movements and the resulting artifacts. In contrast to ECG measurement, which records electrical signals very accurately, the radar signal is merely a composition of body-surface movements and can be quickly distorted by other movements, sweat, or clothing [1]. This is supported by the fact that the measured heart rate was higher than with the reference method. A positive finding was that wearing a military Kevlar protective vest had no significant influence on the heart rate measurements by the radar system.

Correlation of breathing rate

In contrast, there was only a weak, statistically significant correlation for the breathing rate for the condition without the protective vest, and none for wearing the protective vest. This indicates a greater influence of the protective vest on the accuracy of the result.

The radar system also tended to measure higher respiratory rates than the reference system. This again suggests a systematic overestimation of the respiratory rate by the radar system, for example, due to the greater complexity of the respiratory movements or possible interference from environmental factors. An inadequate fit of the breathing belt could also have produced deviations. Comparable FMCW radar studies with a subject-to-radar distance of 1–2 m measured lower deviations in respiratory rate [14].

Effect of the protective vest

Wearing the protective vest had a negative effect on the correlation, particularly for the respiratory rate measurement. For heart rate, the correlation decreased when the vest was worn. This could be due to the protective vest attenuating the signals detected by the radar, or to a restriction in breathing depth, which led to poorer signal quality due to a reduced breathing excursion. Wearing the protective vest may also have caused a change in breathing technique, which was noticeable in altered detectability by the radar system. However, it is striking that the measured values’ scattering range hardly changed.

The measurement position clearly influenced the results for both heart rate and respiratory rate. Without a protective vest, the highest correlations between the radar system and the reference measurement were found in the supine position. This indicates clear detectability of body-surface movement in this position, for example, due to a lower artifact rate. In contrast, the prone position showed the lowest correlation. The reason for this could be a reduced range of movement, for example, due to background pressure on the thorax and abdomen, or interference from the proximity of the ventral thoracic wall to the background. For the condition with the protective vest, the highest correlation was observed for heart rate in the left-side position, whereas the right-side position showed the lowest correlation. This correlates with the asymmetrical positioning of the heart in the prone position, which is made more difficult by the measurement conditions imposed by the vest. The supine position showed both the strongest and weakest correlations for respiratory rate. The good correlation could again be explained by the easier-to-detect signal with ventral positioning; the lowest correlation is probably due to the signal attenuating effect of the protective vest. The literature also shows heterogeneous results regarding the measurement position, with and without an influence of position [11][14].

Strengths and limitations

The accurate detection of heartbeat and respiratory rates by radar systems is primarily impaired by background noise, such as environmental vibrations and electromagnetic interference [3], as well as involuntary body movements (random body movements [RBMs]). Background noise (e.g., electrical devices) can interfere with vital-parameter signals, while RBMs (e.g., coughing, minimal muscle twitches, weight shifts) typically generate nonlinear phase shifts that make it difficult to separate signal components [5][6]. This could have had an influence on the accuracy of our results. The present study has strengths and limitations. The strengths of the present study include the different positions of each subject and the use of a military protective vest for this particular mission. To the best of our knowledge, this is the first study to examine measurement accuracy with and without a protective vest in this setting. By considering a larger number of different positions, the influence of these on the radar measurement could also be shown. The approach demonstrates the potential for contactless measurement of respiratory and heart rates at relatively large distances, which could be leveraged in the future for integration on a drone.

In contrast, a rather small sample size of exclusively cardiopulmonary-healthy subjects and a largely male gender distribution was examined. A larger and more diverse group of test subjects would certainly increase the statistical significance of the present study. Second, the basic signal-processing chain is sensitive to noise, which affects measurement accuracy. More advanced approaches are required to improve the reliability of the measurements. Furthermore, the measurement time per position had to be shortened by 20 seconds at the start of the measurement to rule out a radar calibration problem, and a single value had to be calculated from the remaining 40 seconds.

Due to the underlying mechanism, (un)conscious movements of the test person or other artifacts may have led to the radar system detecting higher heart activity than the ECG-based reference system. This could have influenced the representativeness of the results, as measurement inaccuracies and physiological fluctuations have a greater impact. An inadequate fit of the breathing belt can also influence the results, as it would invalidate the reference measurement. The high failure rate of the reference system during this measurement actually supports this conclusion.

Likewise, possible interference from environmental factors such as the subjects’ choice of clothing, electromagnetic interference or ground conditions cannot be ruled out. The latter was reduced as far as possible by always using the same test conditions in a test setting. With the exception of the protective vest, the choice of clothing was not checked, as the clothing of possibly injured or wounded persons in a crisis or disaster area cannot be influenced in a real operation either. To address these limitations, future studies should ensure closer monitoring of the breathing belt and systematically check for potential failures in the reference system. Alternatively, other methods for valid and reliable measurement of respiratory rate should be used. Longer measurement times and a larger, more diverse sample could help to increase the statistical significance. Further research in signal processing and filtering can minimize artifacts and improve the accuracy of the results.

Conclusions

Despite the limitations and sources of error noted, this study demonstrates the clear potential of the contactless radar system for various application scenarios. In particular, radar measurement offers advantages in the field of emergency medicine in a military setting or in medical disaster operations, especially if it could be used quickly and airborne with a suitable drone.

Funding: The Fraunhofer Institutes FHR and IPA were funded by the German Federal Ministry of Defense for this study.

Conflicts of Interest: KK, DS, SS are active Bundes­wehr officers and work for the Federal Ministry of Defence. DC is an employee of the Federal Ministry of Defence. The Fraunhofer Institutes FHR and IPA were funded by the Federal Ministry of Defense for this study. All authors declare that the research was conducted in the absence of any commercial, financial, or non-financial relationships that could be construed as a potential conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results. This article reflects the opinion of the authors only and not necessarily the opinion of the Federal Ministry of Defense or the Surgeon General of the German Air Force.

References

1. Albrecht NC, Weiland JP, Langer D, Wenzel M, Koelpin A. Characterization of the Influence of Clothing and Other Materials on Human Vital Sign Sensing using mmWave Radar. In: IEEE: 2023 53rd European Microwave Conference (EuMC). Berlin, Germany;2023:428–431. mehr lesen
2. Bhutani A, Marahrens S, Gehringer M, Göttel B, Pauli M, Zwick T. The Role of Millimeter-Waves in the Distance Measurement Accuracy of an FMCW Radar Sensor. Sensors (Basel) 2019;19(18): 3938. mehr lesen
3. Chen KM, Huang Y, Zhang J, Norman A. Microwave life-detection systems for searching human subjects under earthquake rubble or behind barrier. IEEE Trans Biomed Eng 2000;47:105–114. mehr lesen
4. Guo J, Bian Y, Wang W, Dai H, Chen J. Respiration and heart rates measurement using 77GHz FMCW radar with blind source separation algorithm. Annu Int Conf IEEE Eng Med Biol Soc 2022:842–845. mehr lesen
5. Huan C, Kontou P, Smida SB, Anagnostou DE. Dual-Band CW Radar System for Vital Signs Detection With Interference Cancellation Method. IEEE Trans Microw Theory Tech 2025;73(6):3581-3596. mehr lesen
6. Kebe M, Gadhafi R, Mohammad B, Sanduleanu M, Saleh H, Al-Qutayri M. Human Vital Signs Detection Methods and Potential Using Radars: A Review. Sensors (Basel) 2020;20(5):1454. mehr lesen
7. Kehe K, Steinritz D, Sammito S, Groß B, Sauer S. Translationale Forschung - Vom Schreibtisch auf das Gefechtsfeld: Zur translationalen Forschung im Sanitätsdienst der Bundeswehr. WMM 2024;68(9):381–389. mehr lesen
8. Liebetruth M, Kehe K, Steinritz D, Sammito S. Systematic Literature Review Regarding Heart Rate and Respiratory Rate Measurement by Means of Radar Technology. Sensors (Basel) 2024;24(3):1003. mehr lesen
9. Matsunag D, Izumi S, Okuno K, Kawaguchi H, Yoshimoto M. Non-contact and noise tolerant heart rate monitoring using microwave doppler sensor and range imagery. Annu Int Conf IEEE Eng Med Biol Soc 2015:6118–6121. mehr lesen
10. Rosser JB Jr, Parker BC, Vignesh V. Medical Applications of Drones for Disaster Relief: A Review of the Literature. Surg Technol Int 2018;33(11):17–22. mehr lesen
11. Sacco G, Piuzzi E, Pittella E, Pisa S. An FMCW radar for localization and vital signs measurement for different chest orientations. Sensors (Basel) 2020;20(12):3489. mehr lesen
12. Shi K, Schellenberger S, Will C, Steigleder T, Michler F, Fuchs J, Weigel R, Ostgathe C, Kölpin A. A dataset of radar-recorded heart sounds and vital signs including synchronised reference sensor signals. Sci Data 2020;7(1):50. mehr lesen
13. Texas Instruments. Evaluation module for integrated antenna-on-package (AoP) intel-ligent mmWave sensor [Internet]. TI 2022. [letzter Zugriff: 9. Mai 2026]; verfügbar unter: https://www.ti.com/lit/ug/spruim4b/spruim4b.pdf?ts=1691385128540&ref_url=https %253A %252F %252Fdev.ti.com %252F mehr lesen
14. Wang S, Pohl A, Jaeschke T, Czaplik M, Köny M, Leonhardt S, Pohl N. A novel ultra-wideband 80 GHz FMCW radar system for contactless monitoring of vital signs. Annu Int Conf IEEE Eng Med Biol Soc 2015:4978–4981. mehr lesen