Camera-Based Continuous Heart and Respiration Rate Monitoring in the ICU

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This article evaluates the applicability of a relatively rarely used method for recording vital signs in clinical practice. Specifically, the authors assess the effectiveness of monitoring heart rate and respiratory rate using different types of cameras. Data from 36 intensive care unit patients, with a total duration of 699 hours, were processed and analyzed. Camera-based recordings were captured using RGB and NIR cameras (depending on ambient lighting conditions), focusing on the face and chest. Contact-based measurements, including single-lead ECG, airway flow, pressure, and ECG thoracic impedance recorded by a specialized patient monitor, are used as reference vital sign measurements.  The determination of heart and respiratory rhythms was conducted using Fourier transform. The accuracy metric was approximately twice smaller than the standards for contact-based measurement of these parameters. The authors provide a thorough analysis of recording quality, interruptions, disturbances, and necessary corrections due to factors related to the patients' movement and/or treatment procedures. The results are well-supported and appropriately discussed.  

The abstract correctly present the content of the article. Reference sources are relevant to the content and cited at appropriate places in the text.

There are no significant comments on the content, just one question and editorial note:

Why is the chamber box not positioned above the patient's bed? This would significantly reduce the number of distractions that impede continuous monitoring.

Ln 264: “…below the to be determined …” should be corrected.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors,
The submitted manuscript is very interesting and analytical. Below are my comments that in my point of view could further improve the submitted manuscript.

Comment 1
Introduction line 29. You are refer to the methodology financial cost. Please write a very short paragraph either in the introduction or in the discussion session discussing this specific advantage of the presented method, compared to the others

Comment 2
Introduction line 83, please make a notice for an average BPM in a ICU so as to better understand of the 5 BPM (or the 3 BPM presented in the results) agreement.

Comment 3
Paragraph 2.1, lines 139-140. If the FLIR camera was not used either delete the phrase or comment why you did not use it

Comment 4
Paragraph 2.1, lines 144-145. Please specify (or estimate) a required data storage necessity per person 

Comment 5
Paragraph 2.4. The text "state-of-the-art methods" should be followed by indicative references

Comment 6
Paragraph 2.4.2 line 210. In order to better grasp the application of the describe methodology in clinical practise can you please provide a range of the calculated weights, if this data is extractable.

Comment 7
Paragraph 2.4.3, line 216. I am under the assumption that a BPM as high as 220 is unusual in an ICU, however I may be wrong. Can you please explain the choice of the parameter (i.e. is it just a high enough arbitrary value, or you wish to account for some specific abnormality, or it suggested by literature).

Comment 8
Figure 4. please denote sigma as standard deviation

Comment 9
Figure 8a and 8b line colors, it would be most helpful if you had a persistent color coding to your graphs. For instance blue for RGB, red for NIST in both graphs.

Comment 10
Discussion, If the FLIR camera was used would it produce higher accuracy in your results? Please comment, also see my Comment 3

with kind regards

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

In the manuscript titled “Camera-based Continuous Heart and Respiration Rate Monitoring in the ICU” the authors describe a clinical study comparing the heart rate and respiration rate estimated from a camera-based system to gold standard metrics of the same parameters. The camera-based system demonstrated good agreement with the standard clinical monitoring and presents itself as a reasonable alternative to traditional monitoring. The work is interesting, though I have a few comments to address before publication.

Lines 28 - 31: I feel that the statement that current monitoring is too obtrusive or expensive requires a citation and additional justification. A pulse oximeter, which can be made relatively cheaply, whether providing accurate oximetry metrics or not, could provide the heart rate and respiratory rate and does not have a cost or footprint exceeding that of the camera-based device. As the motivation for the rest of the work, I believe this statement needs to be strengthened.

Line 83: While from context the relevant parameter can be determined, I think distinguishing breaths per minute and beats per minute with different acronyms should be done.

Line 144: Given the size of the hard drive in the recording laptop, can you discuss the data volume in GB/min that you are collecting? It is also not clear if the processing pipeline is being run in real time or what the offline run time of the algorithms is.

Line 187: To highlight the novelty of the work, I think it is essential to distinguish what modifications are applied to the state of the art methods.

Line 201: Do you notice spatial variation in the cardiac phase of the PPG signals across different pixels within the ROI on the face? Other works (https://www.nature.com/articles/srep10494) have demonstrated spatial differences in the phase of the imaging PPG signal, which may result in detrimental effects when averaging across pixels.

Line 211: For the heart rate estimation algorithm, would it be reasonable to add time domain analysis to estimate heart rate to address the effects of arrythmias on the accuracy of the FFT method?

Line 388: It might be nice to include representative signal traces of the PPG and breathing signals from the RGB and NIR cameras.

Line 396: More specific details on why almost half of the cohort was excluded for the ROI brightness investigation could be helpful.

Line 407: Can you discuss the types of arrhythmias seen? Is respiratory sinus arrhythmia enough to disrupt the method or are they more severe arrhythmias?

Line 419: How often does the initial 1 frame/minute screening catch or miss segments where the disruption to the signal is significant?

Line 562: Are the distribution of irregular heart rate measurements equally split between post-operative and emergency patients?  

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors thank you for addressing my comments satisfactory. I did find however a typo. Although your answer indicates that the FLIR sentence has been deleted, in my version of the downloaded paper the text "The thermal camera (FLIR Lepton 140 PureThermal 2) ran at 8 fps and 160x120 resolution but was unused in this study." is still visible (see lines 140-141).

Please delete it so as to be in accordance with your review answer

kind regards 

Author Response

Thank you very much again for taking the time to review the manuscript. Please find our response to your comment below. Any changes to the manuscript have been marked in red. 

Comments 1: Dear authors thank you for addressing my comments satisfactory. I did find however a typo. Although your answer indicates that the FLIR sentence has been deleted, in my version of the downloaded paper the text "The thermal camera (FLIR Lepton 140 PureThermal 2) ran at 8 fps and 160x120 resolution but was unused in this study." is still visible (see lines 140-141).

Please delete it so as to be in accordance with your review answer

Response 1: Thank you very much for your comment and careful review of the text. We have now correctly removed the sentence on line 140-141 from the manuscript. We also double-checked to ensure no other references were missed and found one reference on line 131 and therefore we have altered the sentence on line 131 to now read:

"Video recording utilized two camera types: RGB and NIR.".