Non-Contact Heart Rate Detection Based on Hand Vein Transillumination Imaging

Round 1

Reviewer 1 Report

Very interesting paper, with a vein image pattern capturing method not seen before. Some minor corrections/additions:

1. section 3.1, current is given in MA, that should likely be mA if you take standard 850 nm LEDs.
2. The filter mentioned "HS-850LGP-F4.5" cannot be found by internet search. Would be better to mention manufacturer and type separately.

 

Author Response

Response to Reviewer 1 Comments

 

Thank you very much for your review of our manuscript and your valuable comments. We have made corresponding modifications to the manuscript according to your comments. The following is our responses to your suggestions, which has been reflected in the revised paper.

 

Very interesting paper, with a vein image pattern capturing method not seen before. Some minor corrections/additions:

Response:

Thank you for your approval. We have responded to your comments one by one, and revised accordingly in the original manuscript.

 

Point 1: section 3.1, current is given in MA, that should likely be mA if you take standard 850 nm LEDs.

Response 1:

Thank you for your correction, the current unit should indeed be mA, and we have modified the corresponding content in the article.

Corresponding content in the paper:

Among them, the infrared light-emitting tube light source used in the experiment has an emission angle of 45 degrees, a voltage of 1.3-1.6V, a current of 20-30mA, and a peak wavelength of 850nm.

 

Point 2: The filter mentioned "HS-850LGP-F4.5" cannot be found by internet search. Would be better to mention manufacturer and type separately.

Response 2:

It is true that the manufacturer of the filter cannot be effectively known directly through HS-850LGP-F4.5, so we have added the manufacturer of the filter (manufacturer: Hua Shang Laser) according to your suggestion and reflected it in the original manuscript.

Corresponding content in the paper:

It consists of an infrared light-emitting tube, a 850nm narrowband filter (manufacturer: Hua Shang Laser, model: HS-850LGP-F4.5), a 150-degree fisheye lens camera (model: 3200_720P) , and a computer (model: HP Pavilion Gaming Desktop 690-05xx), and MATLAB software (MATLAB version: R2018a) which is further employed to process image.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper entitled “Non-contact Heart Rate Detection Based on Hand Vein Transillumination Imaging” reports about a method of heart rate detection from transillumination imaging.  The topic is interesting, however some concerns need to be addressed:

MAJORS

• The paper is difficult to follow. Please consider organizing the paper in the canonical structure: Introduction, Materials and Methods, Results and Discussion. For instance, the qualitative index and quantitative index are introduced in the results and discussion, whereas they should be introduced in the Methods section.
• Concerning the Bland-Altman plot, I suppose that 100 points should be displayed in the plot, but, as reported in line 300 pag.8 they are 65 points. Please justify why these data are missing. Moreover, no information regarding the heart rate acquisition is reported (e.g. the ECG system, the preprocessing and heart rate detection for ECG signal).
• The Authors stated that their results are acceptable (line 305 pag. 9), or that their indices are below the acceptable threshold (line 360, pag. 10). However, they do not report references of these thresholds or a justification to consider them acceptable. Please provide this information.
• The Authors used a light source that is hand held by the participants (Figure 2). The light propagates in the biological tissue and it is then collected by the camera. The light, during this propagation, crosses also arteries (not only veins), whose motility is more influenced by the pressure wave from the heart. Hence, arteries provide more information regarding the heart rate. In fact, PPG is commonly used to assess the arterial stiffness. Why do the Authors state that they consider the veins to estimate the heart rate? Is the arterial contribution discarded? Please specify this aspect. Please refer to:
  • Dall’Olio, L., Curti, N., Remondini, D., Harb, Y. S., Asselbergs, F. W., Castellani, G., & Uh, H. W. (2020). Prediction of vascular aging based on smartphone acquired PPG signals. Scientific reports, 10(1), 1-10.

 

MINORS

• Some spelling errors are present in the manuscript, for instance some capital letters are present without the a dot before.
• Line 218, pag. 6. The Authors stated that they use 300 frames for their data processing. However, 20 s of acquisition with 20 Hz of sampling frequency, we have 400 frames. Please specify why some frames were not considered.
• Line 187 pag. 5. Please check if it is 20-30 MA or mA
• The Authors stated to use this method to assess the human emotional state. This could be a very interesting application of PPG to be employed, for instance, in clinical facilities to monitor the emotional conditions of the patients. Please refer to:
  • Perpetuini, D., Chiarelli, A. M., Cardone, D., Filippini, C., Rinella, S., Massimino, S., ... & Merla, A. (2021). Prediction of state anxiety by machine learning applied to photoplethysmography data. PeerJ, 9, e10448.

Author Response

Response to Reviewer 2 Comments

 

 

MAJOR

The paper entitled “Non-contact Heart Rate Detection Based on Hand Vein Transillumination Imaging” reports about a method of heart rate detection from transillumination imaging.  The topic is interesting, however some concerns need to be addressed:

Response:

Thank you for your approval. We have responded to your comments one by one, and revised accordingly in the original manuscript.

 

 

Point 1:The paper is difficult to follow. Please consider organizing the paper in the canonical structure: Introduction, Materials and Methods, Results and Discussion. For instance, the qualitative index and quantitative index are introduced in the results and discussion, whereas they should be introduced in the Methods section.

Response 1:

Thank you for your suggestion. We have revised the structure of the paper as a whole, reorganized the paper according to the standard structure you suggested, and introduced the qualitative and quantitative index principles involved in the results and discussions into the method part, thereby simplifying the results and the content of the discussion, in turn, make the paper more convenient for readers to read.

Corresponding content in the paper:

1. Introduction

2.Materials and methods

2.1. Acquisition of vein transmission images

2.2. Acquisition and pre-processing of one-dimensional pulse wave signal

2.2.1.Acquisition of one-dimensional pulse wave signal

2.2.2.Elimination of baseline drift based on morphological operation

2.3. FFT-based heart rate calculation

2.4. Performance evaluation of heart rate detection algorithm

2.4.1. Qualitative index

2.4.2. Quantitative index

3. Results and Discussion

3.1. Qualitative index

3.2. Quantitative index

4. Conclusions

 

 

Point 2: Concerning the Bland-Altman plot, I suppose that 100 points should be displayed in the plot, but, as reported in line 300 pag.8 they are 65 points. Please justify why these data are missing. Moreover, no information regarding the heart rate acquisition is reported (e.g. the ECG system, the preprocessing and heart rate detection for ECG signal).

Response 2:

In response to the lack of data that you raised: Thank you for your detailed opinion. We have checked your opinion. It is because the analysis of our experimental results has made an error. The repeated data has been ignored, which led to the error in the statistics of the data. We have modified this, and the correct value should be 7/100.

In response to your statement that there is no report on the information about heart rate collection, it may be related to the organizational structure of the paper proposed in your first suggestion. Due to our irregular paper structure, you cannot clearly see the introduction of our team's heart rate collection information. We have revised the structure of the paper as a whole according to your suggestions. The heart rate acquisition device and process are introduced in the section "2.1. Acquisition of vein transmission images" in "2. Materials and methods" of this version of the paper.

Corresponding content in the paper:

It can be seen from Figure 7(b) that the 95% consistency is limited to -5.3~5.8 bpm, and 7.00% (7/100) points are outside the 95% consistency limit.

 

2.Materials and methods

2.1. Acquisition of vein transmission images

 

 

Point 3: The Authors stated that their results are acceptable (line 305 pag. 9), or that their indices are below the acceptable threshold (line 360, pag. 10). However, they do not report references of these thresholds or a justification to consider them acceptable. Please provide this information.

Response 3:

Thank you for your suggestion. We have added relevant evidence to support our conclusion and reflected it in the original text.

In the experimental data we have measured, the difference between the camera measurement and the true heart rate is up to 7 bpm, and the average error is 2.28 bpm. According to the pharmaceutical industry standard of the People's Republic of China (error<=5bpm), it can be seen that the error is acceptable. We added this basis to the paper to make the conclusion more evidence-based.

Through the analysis of our experimental data, the values of M, SD, and RMSE are 2.2814, 1.1373, and 2.8254, respectively. By referring to relevant literature, we can know that when the value of the index is controlled at about 4 or less, it can better illustrate the feasibility of the algorithm for heart rate detection . And when the value of the index is close to or greater than 10, it indicates that the algorithm is in a completely invalid state for heart rate detection.We have added relevant papers in the paper to support our conclusion.

Corresponding content in the paper:

The maximum value of the absolute value of the difference is 7bpm, the average of the error is 2.28 bpm. According to the pharmaceutical industry standard of the People's Republic of China(error<=5bpm), it can be seen that the error is acceptable.

Based on the work [20], we can see that when the value of the index is controlled at about 4 or less, it can better illustrate the feasibility of the algorithm for heart rate detection. And when the value of the index is close to or greater than 10, it indicates that the algorithm is in a completely invalid state for heart rate detection.

[20] Li Q, Roger G M, Gari D C, et al. Heart Rate Estimation Algorithm Based on Signal Quality Estimation andKalman Filter [J]. Chinese Journal of Medical Physics, 2007(06):454-457+453.

 

 

Point 4: Dall’Olio, L., Curti, N., Remondini, D., Harb, Y. S., Asselbergs, F. W., Castellani, G., & Uh, H. W. (2020). Prediction of vascular aging based on smartphone acquired PPG signals. Scientific reports, 10(1), 1-10. • The Authors used a light source that is hand held by the participants (Figure 2). The light propagates in the biological tissue and it is then collected by the camera. The light, during this propagation, crosses also arteries (not only veins), whose motility is more influenced by the pressure wave from the heart. Hence, arteries provide more information regarding the heart rate. In fact, PPG is commonly used to assess the arterial stiffness. Why do the Authors state that they consider the veins to estimate the heart rate? Is the arterial contribution discarded? Please specify this aspect. Please refer to:

Dall’Olio, L., Curti, N., Remondini, D., Harb, Y. S., Asselbergs, F. W., Castellani, G., & Uh, H. W. (2020). Prediction of vascular aging based on smartphone acquired PPG signals. Scientific reports, 10(1), 1-10.

Response 4:

Thanks for your comments. Our team has been engaged in the research of vein recognition technology for a long time and has published many related works. In order to prevent others from forging vein features to attack the vein recognition system, this work uses heartbeat detection to improve the security performance of the vein recognition system. Based on this background, this paper proposes a heart rate detection method based on dorsal hand vein images.

In the current vein recognition, because the vein is located on the surface of the skin, when the finger is irradiated with 850 nanometer near-infrared light, the vein characteristic image can be obtained, and then the characteristic image of vein can be extracted and recognized to realize identity authentication. The arteries are located deep in the skin, and it is difficult to collect the characteristic images of the arteries with this imaging device, so this article uses vein images to estimate the heart rate.

Your opinion is very good. In the future, we will develop some sensors to obtain arterial feature images to achieve heart rate estimation and identity recognition.

 

 

MINOR

Point 1: Some spelling errors are present in the manuscript, for instance some capital letters are present without the a dot before.

Response 1:

Thank you for your comments. We have checked and corrected the spelling of the full text. The specific changes are as follows

Corresponding content in the paper:

• Within the consistency limit, the true heart rate is compared with the measured value by camera.
• In the experiment, using near-infrared as the light source, the light source transilluminates the hand, and the transillumination images are collected by a common camera.
• Heart rate is an indicator of the physical health of human body and has important applications in medical field, emotional analysis and other fields.
• However, because they mainly rely on external objects, once the identification items and identification knowledge that prove the identity are stolen or forgotten, its identity is easily impersonated or replaced by others.
• The vein is hidden in the body, it is an inherent feature that is difficult to forge and steal.
• If the heartbeat value is not within the normal range [3,4], it may be intimidated or extremely irritated, so that even oneself cannot perform normal authentication, thus ensuring the security of
• The near-infrared light was used as the light source, and fisheye lens camera was used to obtain high-information images, and then further extracting images.
• M refers to the average of the heart rate measurement error values, which reflects the deviation of the measurement result from the reference heart rate. It is negatively related to the accuracy of the heart rate detection, the calculation formula is shown in Equation (6).
• The higher the true heart rate is with the measured value obtained by camera, the greater the line representing the average difference (the blue solid line, that is, the average of the HR difference). It is close to the line (orange dashed line) representing the average of the difference being 0, which can also reflect the higher consistency between the two.
• Common heart rate monitoring methods include electrocardiographic signal method, arterial blood pressure method, photoplethysmography (PPG) and imaging photoplethysmograph (iPPG) and so on.

 

 

Point 2: Line 218, pag. 6. The Authors stated that they use 300 frames for their data processing. However, 20 s of acquisition with 20 Hz of sampling frequency, we have 400 frames. Please specify why some frames were not considered.

Response 2:

The data obtained in the initial stage of collection may be unstable, so we often use the data collected in the intermediate stage for later experimental processing and analysis. According to your question, in order to make the content of the manuscript more conducive to readers' understanding, we have supplemented the reasons for selecting some frame images in the corresponding part of the original manuscript.

Corresponding content in the paper:

Obtain 20s video, get 400 frames of images by extracting video frames. Since the data obtained in the initial stage of acquisition may be unstable, 300 frames of the intermediate acquisition stage are selected for experimental data processing and analysis, one of which is shown in Figure 3(a).

 

 

Point 3: Line 187 pag. 5. Please check if it is 20-30 MA or mA

Response 3:

Thank you for your correction. This is our spelling error, it should indeed be mA. We have correct the original content.

Corresponding content in the paper:

Among them, the infrared light-emitting tube light source used in the experiment has an emission angle of 45 degrees, a voltage of 1.3-1.6V, a current of 20-30mA, and a peak wavelength of 850nm.

 

 

Point 4: The Authors stated to use this method to assess the human emotional state. This could be a very interesting application of PPG to be employed, for instance, in clinical facilities to monitor the emotional conditions of the patients. Please refer to:

Perpetuini, D., Chiarelli, A. M., Cardone, D., Filippini, C., Rinella, S., Massimino, S., ... & Merla, A. (2021). Prediction of state anxiety by machine learning applied to photoplethysmography data. PeerJ, 9, e10448.

Response 4:

Thank you very much for this reference. We have read through this article. This article studies the ability of a multivariate data-driven method to estimate the state anxiety of healthy participants from the PPG characteristics obtained from the brachial and radial arteries. The principle is that it is affected by the psychophysiological state. PPG may encode information about emotional conditions. Information. Our team also has a deep interest in using the characteristics of PPG signals to study emotions. Related research is currently underway. Thank you very much for your recognition of this research direction.

 

 

 

 

 

 

 

Author Response File: Author Response.pdf

Reviewer 3 Report

The paper presents a non-contact heart rate detection solution using near-infrared images. The idea is interesting, and the potential applicability is undeniable. The system was tested with real data, and, according to the authors, the results prove that the detection accuracy is acceptable. I have some comments regarding the presentation of the work:

1. There are several English mistakes and strange phrase constructions, making the paper difficult to read. Furthermore, there are phrases that do not start with uppercase letter. There are also words in the middle of a sentence, that should start with a lowercase letter, but start with an uppercase one. I point out some of these mistakes. In my opinion, the authors should proofread the paper (asking a professional editor or a native English speaker):
   • “If the heartbeat value is not within the normal range [3,4], it may be intimidated or extremely irritated, so that even oneself cannot perform normal authentication, thus ensuring the security of the system”
   • “This method provides information for judging whether it is a living body for vein authentication, that is, heart rate information, which has great practical application value.”
   • “Because of the flow of blood in the arteries, the absorption of light Nature has also changed.”
   • “The signal is transformed from time domain to frequency domain through FFT, and then study the frequency spectrum structure and change law of signal.”
   • “For example, the width of the structural element is, and the width of …”
   • “Through the peak-to-peak value detection to determine the full-period pulse wave signal, and then…”
2. Regarding the presentation of the method, I do not understand how the authors determine the region of interest from the original image. Is it manually extracted? Or is there an algorithm to determine it? Also, the authors rely too much on the assumption that the users do not move their hand at all during the scan. They should address this problem, and maybe determine the region of interest in each frame, considering the movement of the user.
3. I also have some reservations regarding the complexity of the method. The presented steps, of summing up (and averaging) the intensities of the grayscale information at each frame and transforming the signal into a pulse, performing morphological operations and FFT, are all simple operations. Also, some of these operations are similar to steps from other papers (e.g., Sani et al. “Determination of heart rate from photoplethysmogram using Fast Fourier Transform”). The authors should highlight their original contributions to the domain, and the complexity of their solution.
4. Regarding the evaluation, the authors should explain how they obtained the true heart rates, to compare against the results of their system. Also, the authors state that “The r value is 7.5223, which is much higher than the acceptable threshold”. What is the acceptable threshold? How was this threshold established?

Author Response

Response to Reviewer 3 Comments

The paper presents a non-contact heart rate detection solution using near-infrared images. The idea is interesting, and the potential applicability is undeniable. The system was tested with real data, and, according to the authors, the results prove that the detection accuracy is acceptable. I have some comments regarding the presentation of the work:

Point 1: There are several English mistakes and strange phrase constructions, making the paper difficult to read. Furthermore, there are phrases that do not start with uppercase letter. There are also words in the middle of a sentence, that should start with a lowercase letter, but start with an uppercase one. I point out some of these mistakes. In my opinion, the authors should proofread the paper (asking a professional editor or a native English speaker):

• “If the heartbeat value is not within the normal range [3,4], it may be intimidated or extremely irritated, so that even oneself cannot perform normal authentication, thus ensuring the security of the system”
• “This method provides information for judging whether it is a living body for vein authentication, that is, heart rate information, which has great practical application value.”
• “Because of the flow of blood in the arteries, the absorption of light Nature has also changed.”
• “The signal is transformed from time domain to frequency domain through FFT, and then study the frequency spectrum structure and change law of signal.”
• “For example, the width of the structural element is, and the width of …”
• “Through the peak-to-peak value detection to determine the full-period pulse wave signal, and then…”

Response 1:

Thank you for your careful suggestions. We have made corresponding corrections according to your suggestions, and then revised the grammar of the full manuscript. On this basis, we have sought the help of the polishing agency, hoping that the current version of the paper is acceptable.

 

Point 2: Regarding the presentation of the method, I do not understand how the authors determine the region of interest from the original image. Is it manually extracted? Or is there an algorithm to determine it? Also, the authors rely too much on the assumption that the users do not move their hand at all during the scan. They should address this problem, and maybe determine the region of interest in each frame, considering the movement of the user.

Response 2:

Thank you for your advice. In this article, we subjectively identified region of interest. At present, there are indeed many research teams specializing in the determination of the region of interest. In this paper, we focus more on the verification of using this method to provide living information for vein authentication, so we do not focus on the study of the region of interest algorithm.

In addition, it is true that what you said is an inevitable problem in the experiment, and we cannot guarantee that the subjects are completely in a static state during the experiment. Thank you very much for your suggestion. Since the idea is in the preliminary stage at present, there are some imperfections in many aspects. This paper verifies the feasibility of our idea. In future studies, we will further consider all details of the experiment and improve the experiment and subsequent processing from multiple aspects.

 

Point 3: I also have some reservations regarding the complexity of the method. The presented steps, of summing up (and averaging) the intensities of the grayscale information at each frame and transforming the signal into a pulse, performing morphological operations and FFT, are all simple operations. Also, some of these operations are similar to steps from other papers (e.g., Sani et al. “Determination of heart rate from photoplethysmogram using Fast Fourier Transform”). The authors should highlight their original contributions to the domain, and the complexity of their solution.

Response 3:

Indeed, we did not propose complex processing methods, and the experimental data processing part adopted relatively simple steps. However, we would like to clarify that the introduction of heart rate into venous authentication to improve safety and the experimental equipment involved in this paper are our innovation point, and the subsequent experimental steps are the authentication of our ideas rather than the focus of the article. Although the method involved in the experimental processing part is relatively simple, it is enough to verify our idea. Meanwhile, thank you very much for your opinion, which is also a direction we will explore in the future research.

 

Point 4: Regarding the evaluation, the authors should explain how they obtained the true heart rates, to compare against the results of their system. Also, the authors state that “The r value is 7.5223, which is much higher than the acceptable threshold”. What is the acceptable threshold? How was this threshold established?

Response 4:

In "2.1. Acquisition of Vein Transmission Images" under "2.Materials and Methods", we introduce the acquisition method of truth data and regard the value obtained from contact measurement (sphygmomanometer) as true value. The results are compared with the data obtained in this paper.

Thank you for your suggestion. We have added relevant evidence to support our conclusion and reflected it in the manuscript. In the experimental data we have measured, the difference between the camera measurement and the true heart rate is up to 7 bpm, and the average error is 2.28 bpm. According to the pharmaceutical industry standard of the People's Republic of China (error<=5bpm), it can be seen that the error is acceptable. We added this basis to the paper to make the conclusion more evidence-based.

Through the analysis of our experimental data, the values of M, SD, and RMSE are 2.2814, 1.1373, and 2.8254, respectively. By referring to relevant literature, we can know that when the value of the index is controlled at about 4 or less, it can better illustrate the feasibility of the algorithm for heart rate detection. And when the value of the index is close to or greater than 10, it indicates that the algorithm is in a completely invalid state for heart rate detection. We have added relevant papers in the paper to support our conclusion.

Author Response File: Author Response.doc

Reviewer 4 Report

This manuscript describes the application of a Heart Rate (HR) detection method on videos from image-based Photoplethysmography (PPG). This application is proposed to increase the security of biometrics technologies, by identifying a fake hand manufacture or to assess the emotional status of the person.

The application is interesting to be reported, but there are some major issues that need to be fixed before publication.

Major issues

1. The Author should better organize the paper.
- The description of quantitative and qualitative indices they used to evaluate the method should be included in a “Materials and Methods” section. Probably, sections 2, 3, 4.1, 4.2 should be merged together.

2. Quantitative and Qualitative indices
- For both the Quantitative and Qualitative indices, the Authors state that the results were acceptable, but they did not motivate this statement. How did they assess the acceptability thresholds? Are there some reference standard values? This point is very important to allow an objective evaluation of the method.
- The Authors use the Bland-Altman analysis as a quantitative evaluation of the method. However, the Authors should describe earlier how they obtained ground truth HR data. In addition, they should provide some rationale to justify the thresholds they used.
- Just reporting the grand-average, grand-SD, grand-RMSE is not very informative. Instead, the Authors should compute the subject-based indices and then provide some indications about the distribution of such indices across the population.

3. The Authors should make an effort to include more subjects in the experiment

4. The grammar and syntax of the manuscript are inappropriate for a published manuscript. In some cases the reader has to guess what the Authors meant and there are some parts that are very ambiguous. I think this manuscript was written in another language and subsequently translated to English by a third-party. I strongly recommend that the manuscript is fully revised by a proficient English speaker. Moreover, the Authors should make sure that not only “words” are correctly translated but the original meaning of the sentences is translated. I could not understand the whole section 2.3 and point C in section 3.2

Minor issues

1. I am not an expert in biometrics, but the motivation behind this application described in L75-79 seems unrealistic. The assessment of the emotional status seems more interesting. However, the whole part about biometrics (L60-83) lacks support. Authors should provide references to the literature to motivate and justify their statements.

2. The Authors refer to a “grayscale operation” (L 222), but it is not clear what they mean.
3. The Authors should avoid loading the paper with non necessary information. For instance, they can assume the reader knows the theory of the Fast Fourier Transform, or how to compute mean and SD.
4. Pearson correlation index ranges between -1 to 1. Why did the Authors report a correlation index r=7.5223 ?

Author Response

 

Response to Reviewer 4 Comments

 

This manuscript describes the application of a Heart Rate (HR) detection method on videos from image-based Photoplethysmography (PPG). This application is proposed to increase the security of biometrics technologies, by identifying a fake hand manufacture or to assess the emotional status of the person.

The application is interesting to be reported, but there are some major issues that need to be fixed before publication.

Major issues

Point 1: The Author should better organize the paper.
- The description of quantitative and qualitative indices they used to evaluate the method should be included in a “Materials and Methods” section. Probably, sections 2, 3, 4.1, 4.2 should be merged together.

Response 1:

Thank you for your advice, we has modified the overall structure of the paper, according to the standard of your proposed paper structure. At the same time, the principles of qualitative indexes and quantitative indexes in the results and discussion are introduced into the material and method section, which can simplify the content of the results and discussion and make the paper more convenient for readers.

 

Point 2:  Quantitative and Qualitative indices
- For both the Quantitative and Qualitative indices, the Authors state that the results were acceptable, but they did not motivate this statement. How did they assess the acceptability thresholds? Are there some reference standard values? This point is very important to allow an objective evaluation of the method.
- The Authors use the Bland-Altman analysis as a quantitative evaluation of the method. However, the Authors should describe earlier how they obtained ground truth HR data. In addition, they should provide some rationale to justify the thresholds they used.
- Just reporting the grand-average, grand-SD, grand-RMSE is not very informative. Instead, the Authors should compute the subject-based indices and then provide some indications about the distribution of such indices across the population.

Response 2:

Thank you for your suggestion. We have added relevant evidence to support our conclusion and reflected it in the original manuscript. In the experimental data we have measured, the difference between the camera measurement and the true heart rate is up to 7 bpm, and the average error is 2.28 bpm. According to the pharmaceutical industry standard of the People's Republic of China (error<=5bpm), it can be seen that the error is acceptable. We added this basis to the paper to make the conclusion more evidence-based.

Through the analysis of our experimental data, the values of M, SD, and RMSE are 2.2814, 1.1373, and 2.8254, respectively. By referring to relevant literature, we can know that when the value of the index is controlled at about 4 or less, it can better illustrate the feasibility of the algorithm for heart rate detection. And when the value of the index is close to or greater than 10, it indicates that the algorithm is in a completely invalid state for heart rate detection. We have added relevant papers into the manuscript to support our conclusion.

In "2.1. Acquisition of Vein Transmission Images" under "2.Materials and Methods", we introduce the acquisition method of truth data and regard the value obtained from contact measurement (sphygmomanometer) as true value. The results are compared with the data obtained in this paper. According to your last opinion, we have added relevant basis to make the paper more rigorous.

Thank you for your suggestions. In your review comments, you pointed out many imperfections in our current experiment and later analysis. Indeed, the analysis using such indicators as Grand-Average, Grand-SD and grand-RMSE may not be sufficient, but it has already achieved preliminary verification of our ideas. We will try our best to improve all aspects in the future research. Thank you for your good suggestions for our later experimental analysis.

 

Point 3: The Authors should make an effort to include more subjects in the experiment

Response 3:

Thank you for your comments. Although our team has been engaged in vein recognition technology research for a long time, many related work has been published. However, the research direction of trying to introduce heart rate into intravenous authentication to improve system safety is only in the preliminary stage, so there are many imperfections. This paper has verified our idea, and we will further improve the experiment in the future study, including of expanding the number of subjects you mentioned. Thank you very much for your suggestion, which is also the direction of our future efforts.

 

Point 4: The grammar and syntax of the manuscript are inappropriate for a published manuscript. In some cases the reader has to guess what the Authors meant and there are some parts that are very ambiguous. I think this manuscript was written in another language and subsequently translated to English by a third-party. I strongly recommend that the manuscript is fully revised by a proficient English speaker. Moreover, the Authors should make sure that not only “words” are correctly translated but the original meaning of the sentences is translated. I could not understand the whole section 2.3 and point C in section 3.2

Response 4:

Thank you for your advice. It's true that our native language is not English. In order to improve the readability of the article, we sought the help of native English speakers, and further sought the help of polish organization so that readers could better understand the manuscript.

Minor issues

Point 1: I am not an expert in biometrics, but the motivation behind this application described in L75-79 seems unrealistic. The assessment of the emotional status seems more interesting. However, the whole part about biometrics (L60-83) lacks support. Authors should provide references to the literature to motivate and justify their statements.

Response 1:

The motivation behind this application described in L75-79 is realistic, fake hand models do exist to simulate vein features to fool authentication systems，and we added relevant references to the original manuscript to illustrate. These studies have revealed that finger vein biometrics is also vulnerable to presentation attacks [17-21], i.e., printed versions of authorized individual finger vein images can be used to gain access to facilities or services.

 

[17] S. Tirunagari, N. Poh, D. Windridge, A. Iorliam, N. Suki, and A. T. Ho, “Detection of face spoofing using visual dynamics,” IEEE Trans. Inf. Forensics Security, vol. 10, no. 4, pp. 762–777, Apr. 2015.

[18] V. Ruiz-Albacete, P. Tome-Gonzalez, F. Alonso-Fernandez, J. Galbally, J. Fierrez, and J. Ortega-Garcia, “Direct attacks using fake images in iris verification,” in Proc. 1st Eur. Workshop Biometrics Identity Manage. (BioID), vol. 5372. 2008, pp. 181–190.

[19] P. Tome et al., “The 1st competition on counter measures to finger vein spoofing attacks,” in Proc. Int. Conf. Biometrics (ICB), May 2015, pp. 513–518.

[20] G. L. Marcialis et al., “First international fingerprint liveness detection competition–LivDet 2009,” in Proc. ICIAP, 2009, pp. 12–23.

[21] C. Sousedik and C. Busch, “Presentation attack detection methods for fingerprint recognition systems: A survey,” IET Biometrics, vol. 3, no. 4, pp. 219–233, 2014.

 

Point 2:  The Authors refer to a “grayscale operation” (L 222), but it is not clear what they mean.

Response 2:

"Grayscale operation" refers to the process of obtaining grayscale images. The images collected by our camera are color images (three-channel images), and the single-channel images are obtained by grayscale. Your question may also be caused by the unclear expression of our English. I hope the paper can be more understandable after polishing

Point 3: The Authors should avoid loading the paper with non necessary information. For instance, they can assume the reader knows the theory of the Fast Fourier Transform, or how to compute mean and SD.

Response 3:

Thank you for your suggestions. According to your suggestions, we deleted unnecessary information, including the introduction of FFT or average value you mentioned, so as to make the article more concise.

Point 4: Pearson correlation index ranges between -1 to 1. Why did the Authors report a correlation index r=7.5223?

Response 4:

We are sorry for our negligence of details. The correct one should be R =0.7.522. We have corrected it in the paper.

Author Response File: Author Response.doc

Round 2

Reviewer 2 Report

The Authors reply to my concerns in a satisfying manner. In my opinion the paper is suitable for publication.

Reviewer 3 Report

The authors addressed my concerns.

Reviewer 4 Report

Tha authors have done a good work