Selecting Target Range with Accurate Vital Sign Using Spatial Phase Coherency of FMCW Radar

Round 1

Reviewer 1 Report

The paper proposes a method based on the coherency of phase to select with high accuracy the range bin containing accurate vital signs’ information.

1) English should be revised; some sentences are awkward or incorrect.

2) Some additional clarifications about Fig. 1 are necessary.

• How is Fig.1 obtained, did the authors use the proposed algorithm or a solution available in the literature?
• The authors state that the vital signs’ information is contained only in certain range bins. However, it is not clear from the comments how this quantity of information is quantified. A simple observation of the data is not enough to state the presence of the information. Are you processing the data, or the consideration are only based on the signal shape?
• Are the signals of Fig. 1(c) and 1(d) referred to the same person and the same experiment? Typically, the respiratory and heart rate are retrieved in the same range bin. Thanks to the higher resolution in the MMW range it is understandable that the information about the respiratory and heart rates is located in different bins. However, it seems that the bin associated with the respiratory and heart rates are too far apart (till about 20 cm and 40 cm in Fig. 6a) if considering typical body sizes. Please add a more detailed explanation. How do these distances compare with the real distance of the subject from the antenna?

3) Please better clarify the method used to obtain Fig. 2.

4) The signals related to the respiratory and heart rate are selected with two filters (0.1-0.4Hz for respiration, and 0.8-1.7Hz f). Please give further details about the chosen filters.

5) Does the radar used for the measurements contain a PLL?

6) Did the authors tested the performance of the algorithm for higher distances corresponding to a lower signal to noise ratio (SNR)?

Author Response

Please see the attachment. We attached supplementary figures in uploaded WORD file.

- Response #1

Thank you for your advice.
We revised the English of this paper once again through a professional English proofreading institution.

- Response #2 :
To describe more details of our experiment and help understanding signals in Fig.1, we added scene of experimental settings in Fig.1 indicating the distance between subject and radar.

In addition, we additionally referred conventional studies of radar based vital measurements, which state about the body parts for measuring respiration and heartbeat signal [1][2][3]. According to the studies, the respiration can be measured near abdomen or thorax, and heartbeat can be measured where skin is thin, such as neck. Considering conventional studies, respiration signals near 0.6m and heartbeat signals near 0.8m in our experimental settings are reasonable.

In Fig.1, to show that the vital signals exist in specific range bins, we manually searched the range bins where accurate respiration or heartbeat exists compared to the referenced signals. Referenced vital signal measurement device (respiration belt and ECG) and radar are separate, and are not perfectly synchronized. Therefore it is hard to quantify the vital information in the radar using the referenced signal. We showed magnitude and phase, and the extracted respiration and heartbeat signal with the referenced signals in Figure.1. The extracted vital signals are well matched with the referenced vital signals, and the range of respiration and heart are reasonable considering the conventional studies.

In addition, we substituted the expression “vital signs’ information is contained only in certain range bins” to “displacement of vital sign exists in certain range-bins” to avoid confusion.

 

- Response #3 :
We revised the paper to clarify the method used to obtain Fig.2.

“To generate , peak points of sinusoids were detected, the location of peak points were disordered by uniform noise, and the points were interpolated. From  to , the boundary of uniform distribution was expanded to generate further disorder.”

 

- Response #4

Considering that normal adults’ respiration rate is 14~20 RPM and heart rate is 60~100 BPM [4][5], we set 0.1-0.4 Hz filter for respiration and 0.8-1.7 Hz filter for heartbeat with some margin.

We revised the paper to give further details about the filters.
The filters that conventional methods use for respiration and heartbeat are as follows.

Conventional #1 [6] : 0.1-0.5 Hz for respiration, 0.8-2 Hz for heartbeat
Conventional #2 [7] : 0.1-0.4 Hz for respiration, 0.8-1.5 Hz for heartbeat
Conventional #3 [8] : 0.15-0.4 Hz for respiration, 0.83-1.5 Hz for heartbeat

The filters of the proposed method are similar with those of conventional methods.

 

- Response #5

The transceiver chipset of the radar (BTS60 [9]) used in this study is BGT60 of INFINEON company. According to the datasheet of BGT60 [10], the chipset uses ADF4158 circuit for PLL. 

 

- Response #6

We tested performance of the algorithm with regard to various distances only for a subject. As the result shows, the proposed algorithm works well for various distances. However, verification for more subjects is still required to assess the performance. As future works, we will analyze the performance of the proposed algorithm not only for distances, but also posture of subjects or other measurement environments.

 

<Reference>

[1] M. He, Y. Nian, and Y. Gong, “Novel signal processing method for vital sign monitoring using fmcw radar,” Biomedical Signal Processing and Control, vol. 33, pp. 335–345, 2017.

[2] M. Alizadeh, G. Shaker, J. C. M. De Almeida, P. P. Morita, and S. Safavi-Naeini, “Remote monitoring of human vital signs using mm-wave fmcw radar,” IEEE Access, vol. 7, pp. 54 958–54 968, 2019.

[3] J.-Y. Park, Y. Lee, Y.-W. Choi, R. Heo, H.-K. Park, S.-H. Cho, S. H. Cho, and Y.-H. Lim, “Preclinical evaluation of a noncontact simultaneous monitoring method for respiration and carotid pulsation using impulseradio ultra-wideband radar,” Scientific reports, vol. 9, no. 1, pp. 1–12, 2019.

[4] Model, D. Making Sense of Clinical Examination of the Adult Patient: Hands-on Guide; CRC Press, 2006.

[5] Taktak, A.; Ganney, P.; Long, D.; Axell, R. Clinical engineering: a handbook for clinical and biomedical engineers; Academic Press, 2019.

[6] Wang Y, Wang W, Zhou M, Ren A, Tian Z. Remote monitoring of human vital signs based on 77-GHz mm-wave FMCW radar. Sensors. 2020 Jan;20(10):2999.

[7] Anitori L, de Jong A, Nennie F. FMCW radar for life-sign detection. In2009 IEEE Radar Conference 2009 May 4 (pp. 1-6). IEEE.

[8] Sekak F, Zerhouni K, Elbahhar F, Haddad M, Loyez C, Haddadi K. Cyclostationary-Based Vital Signs Detection Using Microwave Radar at 2.5 GHz. Sensors. 2020 Jan;20(12):3396.

[9] BTS60 [accessed on 23 April 2021]; Available online : https://bitsensing.com/60ghz-mini-h/

[10] Infineon Technologies AG Single-Chip SiGe Transceiver Chipset for V-band Backhaul Applications from 57 to 64 GHz, Application note AN 376, Revision: Rev. 1.0. [(accessed on 23 April 2021)]; Available online: https://www.infineon.com/dgdl/Infineon--AN-v01_00-NA.pdf?fileId=5546d4624ad04ef9014aed1c06120a5e.

Author Response File: Author Response.pdf

Reviewer 2 Report

 

This paper presents a method on the selection of range bins in FMCW radar for more accurate vital sign detection. The work is quite interesting with promising results achieved.

 

1. According to the author, 20 subjects were tested in the experimental study, but only results for two subjects were presented and the results were also given in a very short time period (30s). It would be interesting to see the results for a longer time window.  Can these good results be achieved for all the subjects? If not, how the results look like? What are the possible reasons and what a potential solution could be?

 

2. Mathematical expression of mean error, correctness, and correlation coefficient should be given in the paper. Were the values presented in Table 1 obtained for a time window of two minutes? Please specify clearly relevant parameters used for the calculation of these numbers.

 

3. The author stated that “the reflected angle or area of the target varies by delta (t)” and this in consequence could affect the antenna gain, cross section and etc. The reviewer found this statement not solid. How the reflected angle or area of the target would vary when there is a slight chest wall displacement?

 

4. How the t0 in equation (5) was chosen in the experimental study?

Author Response

Please see the attachment. We attached supplementary figures in uploaded WORD file.

 

- Response #1

Thank for your comments.

We attached figures which are the results for 2min whole window. The results for 2min whole window are very dense when conventional methods are included. Also authors have thought that the results for 30s window are not very different from 2min window. For the better presentation of the results, we have decided to highlight the results zoomed in 30s window. The accuracy for subject 1 or 2 is almost the same as those of other subjects and the averaged performance for all 20 subjects is summarized in Figure.6 and Table. 1.

 

- Response #2

We supplemented mathematical expressions of mean error , correctness , and correlation coefficient  in the paper. 

Also, we revised the paper to clarify the calculation of the values in Table. 1.

Using the 15s of time window, vital rates are calculated every second for each subject. Therefore, 105 vital rates are acquired from each subject, so total 2100 vital rates are obtained from 20 subjects. The values presented in Table 1 are obtained using total vital rates of all the subjects.

 

-Response #3

We additionally referred conventional studies about the radar cross section measurements of the human body for breath activity [1],[2]. According to the conventional studies, the radar cross sections of human body at the end of inspiration and expiration differ. Because the wavelength of the radar used in this study is shorter than the conventional studies, slight chest wall displacement would affect the radar cross section, antenna gain, or effective area.

 

- Response #4

The average period of normal adults’ respiration is about 5 seconds [3][4]. To sufficiently utilize the respiration information, we set window time,  as 15 seconds which would include 3 cycles of respiration on average.

 

<Reference>

[1] Piuzzi E, D’Atanasio P, Pisa S, Pittella E, Zambotti A. Complex radar cross section measurements of the human body for breath-activity monitoring applications. IEEE Transactions on Instrumentation and Measurement. 2015 Jan 30;64(8):2247-58.

[2] Aardal Ø, Hamran SE, Berger T, Hammerstad J, Lande TS. Radar cross section of the human heartbeat and respiration in the 500MHz to 3GHz band. In2011 IEEE Radio and Wireless Symposium 2011 Jan 16 (pp. 422-425). IEEE.

[3] Model, D. Making Sense of Clinical Examination of the Adult Patient: Hands-on Guide; CRC Press, 2006.

[4] Taktak, A.; Ganney, P.; Long, D.; Axell, R. Clinical engineering: a handbook for clinical and biomedical engineers; Academic Press, 2019.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The paper was reviewed according to the reviewer suggestions but some concerns about the extraction of the vital signs information need additional work.

1) In the answer to the comment number 2 the authors state that the bins associated with respiration and heartbeat are located at a reasonable distance, but this distance should also correspond to the real distance of the target that should have been measured at the moment of the vital signs acquisition. In addition, you should dispose of a temporal correlation between the vital signs extracted with the radar and the reference to show a fair comparison. Please better discuss this point.

2) Normally the vital signs information corresponds to the respiratory and heart frequencies that clearly emerge in the spectrum of the temporal signals. The authors state that in Fig. 1 there is more signal distortion in certain range bins (r=0.875 m and r=0.65 m). Please clarify what do you mean with this term and better quantify this.

3) In Section 3.3 authors evaluate vital signs accuracy, but in the reality, they are evaluating the ability of the standard algorithm and the proposed one to correctly estimate the range bin inside which the target body element contributing to respiratory / heart rate is located. While the comparison between the two methods is clear, it would be necessary to add some information about how the two algorithms perform if compared with the real distance of the target.

Author Response

Thank you for valuable comments, and an opportunity to address the comments.

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

 

(1) The author should give more explanations on how the range bins were selected in reference [37-40] since the methods were used as benchmarking, especially for reference [38]. According to figure 5, the range bin selected by using the method from reference [38] was about 0.5 m, which is about the half of  the realistic distance 1m. How this range bin was selected and why it is significantly different from the real distance?

(2) According to the author's response to my previous comments, in the 2 min detection data, there are some time moment when the detected respiration rate (between 50s and 60s) and heartbeat rate (between 60s and 70s) have larger deviation from the reference data. Please plot out the detected respiration waveform and heartbeat waveform in comparison with reference data and comment on what are the possible reasons to this bigger deviation.  In addition, I think the author should add the two-minutes window results to the manuscript.

(3) The author said they use 15 s window to calculate the phase coherency. Why this specific number was used? Will the selected range bin be different if the window length is different? Please comment on the possible use of a longer or short time window?

Author Response

Thank you for valuable comments, and an opportunity to address the comments.

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

All the comments/concerns have been answered, I suggest the publication of the manuscript