Effects of Proprioception and Visual Focus of Attention on Surface Electromyography Proportional Control
Round 1
Reviewer 1 Report
GENERAL COMMENTS
The manuscript deals with the effect of different visual feedback conditions on the surface EMG signal used for the reconstruction of elbow angle during vertical and horizontal movements. First of all, some parts are hard to follow and some senteces are not clear. I strongly suggest to revise english in order to improve readability and clarity of concepts.
The aim of the study is well explained in the introduction section and the experimental setup appears appropriately designed. However, I have a couple of concerns regarding the general concept of the manuscript:
- Authors use back propagation NN for elbow angle prediction from sEMG features. However, RMS errors between predicted and actual elbow angles are quite large, showing also a non constant error. Why authors did not test other classification techniques in order to improve their results?
- Another concern regards the errors between predicted and actual angles, that are quite large. Despite I understand that the goal of the manuscript is to provide an evaluation of the influence of different visual attention conditions on the goodness of the elbow angle prediction from sEMG data, the errors appear significant and this aspect is not taken into consideration in the discussion section. Further, authors evaluated also the coherence between EMG and EEG signals, obtaining extremely low values (the highest is 0.2285). Also this aspect has not been discussed adequately and a reader could conclude that the coherence between the two signals is absent, despite it would be anyway significant for the authors aims. These aspects warrant to be carefully discussed into the more general architecture of the manuscript.
- In the discussion section authors in my opinion did not stress adequately their results, presenting also a series of considerations poorly grounded on their results. I would suggest to redraft the discussion section highlighting the aspects directly related with their work and their outcomes.
- Finally, the experimental procedures used for set the different visual attention conditions are not clearly explained. How the proprioceptive, no visual attention and visual attention conditions have been obtained? Given the importance of these aspects for the whole study this issue should be addressed carefully.
SPECIFIC COMMENTS
METHODS
line 96 - please, specify the low-pass filtering cutoff frequency.
lines 155-160 - this part is not clear in my opinion. In what sense the angle recognition accuracy is practical?
lines 235-237 - the description of the internal visual focus and the proprioceptive no visual feedback are not clear. They warrant to be carefully described considering that are a fundamental part of the experimental setup.
lines 258-260 - please, rephrase the sentence, it is not clear in its current form.
lines 261-262 - how did you chose the threshold of 5°? this value appears quite significant for a task with a range between 0° and 90° despite Burdea et al. reported 5° as the upper limit of accuracy of their system. This choice warrants to be carefully justified, considering also that the errors appear with a significant magnitude.
line 266 - please report the measurement unit for the RMSE and also in all the figures.
line 286 - what does it mean "the ontology condition"?
DISCUSSION
line 314 - the elbow movement with external visual focus is more fluent in what terms? how do you quantified it? for instance, from the results, in the EVA condition the error is the biggest.
FIGURES
I would suggest to provide figure with a better resolution: figure 4 and 5 are hard to read.
Author Response
Manuscript ID: applsci-436055
Title: " Effects of Proprioception and Visual Focus of Attention on Surface Electromyography Proportional Control "
Correspondence Author: Wang Weixing
Dear Editors and Reviewers:
We would like to thank Applied Sciences for giving us the opportunity to revise our manuscript.
We thank the Reviewers for their careful read and thoughtful comments on previous manuscript. We have carefully taken their comments into consideration in preparing our revision, which has resulted in a paper that is clearer, more compelling, and broader. The following summarizes how we responded to Reviewer comments. Below is our response to their comments.
Thank you very much for all your help and looking forward to hearing from you soon.
Best regards
Sincerely yours
Wang Weixing
Corresponding Author
Response to Reviewer 1 Comments
GENERAL COMMENTS
Point 1: Authors use back propagation NN for elbow angle prediction from sEMG features. However, RMS errors between predicted and actual elbow angles are quite large, showing also a non constant error. Why authors did not test other classification techniques in order to improve their results?
Response 1: Thanks for the reviewer’s questions. Our goal is to use bioelectrical signals to identify joint angles for proportional control of exoskeletons. As suggested by the reviewer, there are still a large errors in calculating the joint angles using regression method. We believe that there may be many reasons for the error, including the processing method of bioelectric signals, the number of samples of the subjects, the classification regression method, and so on. These are all issues that need to be resolved. Through experiments, we found that the direction of movement, concentration of attention and proprioception also have an impact on the regression results, so we only focus on the analysis of these three aspects. In the processing of experimental data, we also tried the three machine learning methods of support vector machine, BP neural network and random forest, and found that BP neural network has the best results. Because our research focuses on the effects of several psychological conditions, only the calculation results of BP neural network are placed in the manuscript. In future research, in order to reduce the angle recognition error, we also hope to update our biosignal acquisition equipment to obtain sensor array data, expand the amount of data and use deep learning methods.
Point 2: Another concern regards the errors between predicted and actual angles, that are quite large. Despite I understand that the goal of the manuscript is to provide an evaluation of the influence of different visual attention conditions on the goodness of the elbow angle prediction from sEMG data, the errors appear significant and this aspect is not taken into consideration in the discussion section.
Response 2: Thank you very much for this suggestion. We have added a discussion of recognition errors in Section 5 (lines 309-318), as follows: This paper studied the accuracy of angle prediction can be affected by some psychological factors. The basis is that the elbow joint angle can be estimated by the sEMG signal during continuous motion using a classification or regression algorithm. For the elbow joint motion control based on the sEMG signal, at present, the classification recognition error using the motion type or the motion direction is relatively low. However, the features of the sEMG signal are converted to joint angle values through the regression method and then directly applied to the exoskeleton actuator, which is more able to ensure continuous matching of human-machine motion. This also effectively compensates for the inflexibility of classification identification. Although there is still a large error in angular regression, we have found that these errors can be appropriately reduced under some psychological conditions. The results of this paper can be used as a basis for more optimal sEMG control.
Point 3: Authors evaluated also the coherence between EMG and EEG signals, obtaining extremely low values (the highest is 0.2285). Also this aspect has not been discussed adequately and a reader could conclude that the coherence between the two signals is absent, despite it would be anyway significant for the authors aims.
Response 3: Thanks for the reviewer’s suggestion. We have revised the manuscript. In the experimental results, it has been clearly stated that there was no coherence between sEMG and EEG signals under proprioceptive condition, and coherence was observed under both the internal and external conditions (lines 291-293), as follows: Then, in the discussion section, the reason is explained (lines 374-377). As follows: Under the proprioceptive condition, there was almost no coherence between the two signals, and no crosstalk occurs between them. In combination with recognition accuracy, the automatic characteristics of motion and the flexibility of motion were not limited under the proprioceptive condition.
Point 4: The experimental procedures used for set the different visual attention conditions are not clearly explained. How the proprioceptive, no visual attention and visual attention conditions have been obtained? Given the importance of these aspects for the whole study this issue should be addressed carefully.
Response 4: Thanks for the reviewer’s suggestion. We have revised the description for proprioception and attention conditions in the experimental section (lines 226-236). We hope that the key issues of these studies can be explained. As follows: Because the proprioception is the feeling that the moving organs such as muscles, tendons, and joints appear on their own during exercise or at rest. For example, a person can sense the position of various parts of the body when closing his eyes. Therefore, our experiment required that the subject feeled the movement angle of the elbow joint with subjective feeling under the proprioceptive condition, and there was no visual feedback. The subjects were asked to close their eyes slightly and focus on the motion of the elbow joint in heart. The experimental setup of the internal and external attention conditions required visual feedback significantly. In the external focus of attention experiment, the subjects were asked to focus on a display screen (1280 × 800 px) with a front distance of 1 m. The sEMG signal and joint angle waveforms were displayed on the screen and they were the effect of motion. In the internal focus of attention experiment, the subjects were asked to visually notice their elbow joint motions.
SPECIFIC COMMENTS
Point 1: line 96 - please, specify the low-pass filtering cutoff frequency.
Response 1: Thanks for the reviewer’s suggestion. The pre-processing of sEMG signal has been revised in the revision (lines 94-99). As follows: The sEMG signal carries low frequency (near DC) and high frequency interference signals. First, the signal collected from the surface electrode was processed by a high pass filter with a 1 Hz cutoff frequency. Second, the high frequency interference signal was filtered out by a Butterworth low pass filter with a 600 Hz cutoff frequency. Finally, the output signal was processed by a notch filter with a power frequency of 50 Hz as the center frequency to further improve the signal-to-noise ratio.
Point 2: lines 155-160 - this part is not clear in my opinion. In what sense the angle recognition accuracy is practical?
Response 2: Thanks for the reviewer’s suggestion very much. As the reviewer suggested, the choice of threshold (5°) must to be carefully justified. Because the error range lacks sufficient proof and significance. So we revised the reference to Burdea et al., and the relevant comments were also removed from the revision.
Point 3: lines 235-237 - the description of the internal visual focus and the proprioceptive no visual feedback are not clear. They warrant to be carefully described considering that are a fundamental part of the experimental setup.
Response 3: Thanks for the reviewer’s suggestion. We have revised the description for proprioception and attention conditions in the experimental section (lines 226-236). We hope that the key issues of these studies can be explained.
Point 4: lines 258-260 - please, rephrase the sentence, it is not clear in its current form.
Response 4: Thanks for the reviewer’s suggestion. The original sentence has been revised to: It can be seen from the deviation of the regression curve that the recognition error of motion angle under the internal focus of attention condition is the lowest.
Point 5: lines 261-262 - how did you chose the threshold of 5°? this value appears quite significant for a task with a range between 0° and 90° despite Burdea et al. reported 5° as the upper limit of accuracy of their system. This choice warrants to be carefully justified, considering also that the errors appear with a significant magnitude.
Response 5: Thanks for the reviewer’s suggestion very much. As the reviewer suggested, the choice of threshold (5°) must to be carefully justified. Because the error range lacks sufficient proof and significance. So we revised the reference to Burdea et al., and the relevant comments were also removed from the revision.
Point 6: line 266 - please report the measurement unit for the RMSE and also in all the figures.
Response 6: Thanks for the reviewer’s suggestion. In the revision, we have reported the measurement unit for the RMSE (°) and also in all the figures.
Point 7: line 286 - what does it mean "the ontology condition"?
Response 7: Thanks for the reviewer’s suggestion. We have revised this typo, and this should be "the proprioceptive condition".
Point 8: line 314 - the elbow movement with external visual focus is more fluent in what terms? how do you quantified it? for instance, from the results, in the EVA condition the error is the biggest.
Response 8: Thanks for the reviewer’s suggestion. External visual focus has a lower RMSE value than internal visual focus. This showed that the accuracy of elbow joint angle is higher under external visual focus condition. We think this may be because external visual focus of attention reduces the amount of attention required for action operations, reducing the cost of the task at the cognitive level. In contrast, the limitations of internal visual focus of attention not only come from the subject's movements themselves, but also the interaction of other muscles, even the entire motor system.
Point 9: I would suggest to provide figure with a better resolution: figure 4 and 5 are hard to read.
Response 9: Thanks for the reviewer’s suggestion. We have carefully revised Figures 4 and 5 for better reading.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The work is compelling and appears to have been well thought out and well conducted. The effect of proprioception on performance is a fascinating topic. I was eager to read this paper.
The difficulty for me lies in a bit of confusion. I am not clear whether I am misunderstanding the experiment, I am misunderstanding the results, or I am misunderstanding the authors' interpretation of the data.
I hope that the error lies in the former. Lines 227-238 describing the EVA, IVA, and NVF psychological factors were difficult to understand. I believe I understand EVA and IVA. However, NVF is still confusing to me. Were subjects asked to close their eyes and move forearm by 'feel'?
This should be described more thoroughly.
Assuming I have understood this portion, it is quite surprising that data showed NVF (eyes closed, move by feel) yielded the best data (lowest overall error (Fig6) and low local error (figs4, 5).
If I am understanding the experiment and data correctly, Discussion lines 302-304 are not supported by data. Accuracy is not improved by external visual data. In both cases, accuracy was improved by closing the eyes.
I hope I am misunderstanding the test setup.
Some other concerns about the paper (not the experiments):
The figure layout needs a great deal of work:
- Text font,and size varies throughout the document. Text in Figs 4, 5 are so small as to be illegible. Returning to the electronic copy and zooming in, the text is so pixelated as to be illegible. It is difficult to review a paper when the main results figures are illegible.
- Figures are poorly thought out. Figure2, a great deal of space is given to the device, though the lower half of the image only shows the wearer's legs and chair. Figure 3 is quite large (could easily shrink to 1/2 the height but text is too small even at this size. Figures4, 5 need to be completely redone. This is the heart of the paper, and it's difficult to tell what's going on. Why is green shaded in figure 5? Isn't that deviation from nominal? Then isn't distance from the red line the key aspect? Is the redline on zero degrees? can't read it. Figure 6: the only data is the six values. that could easily be in table or even text. The bar chart is large and conveys little information. Perhaps just an additional column in Table 3. Figure 7. it was never made clear the importance of Coherence. Probably eliminate it. If it remains, chop the top half. there is no data there.
- Section 2 can be greatly reduced. Sections 2.2, 2.3, 2.4 describe statistical methods in EMG. This is not novel and the authors do not claim that it is. Simply cite and explain in a few sentences. Section 2.5 should either be re-worked to justify Figure 7, or both can be eliminated.
- It is difficult to comment on discussion and conclusions as the method was not clear. If my interpretation above is correct, then discussion is inaccurate. I hope I am simply misunderstanding lines 227-238.
Overall, this work is quite compelling. I hope that with some clarification, the procedure, data, and discussion match up, as this result could be quite interesting.
Please work on the figures. They need quite a bit of work. I assume formatting was left for a future submission, but better figures would be very welcome. Figures 4 and 5 (the main data) were un-readable.
Author Response
Manuscript ID: applsci-436055
Title: " Effects of Proprioception and Visual Focus of Attention on Surface Electromyography Proportional Control "
Correspondence Author: Wang Weixing
Dear Editors and Reviewers:
We would like to thank Applied Sciences for giving us the opportunity to revise our manuscript.
We thank the Reviewers for their careful read and thoughtful comments on previous manuscript. We have carefully taken their comments into consideration in preparing our revision, which has resulted in a paper that is clearer, more compelling, and broader. The following summarizes how we responded to Reviewer comments. Below is our response to their comments.
Thank you very much for all your help and looking forward to hearing from you soon.
Best regards
Sincerely yours
Wang Weixing
Corresponding Author
Response to Reviewer 2 Comments
Point 1: Lines 227-238 describing the EVA, IVA, and NVF psychological factors were difficult to understand. I believe I understand EVA and IVA. However, NVF is still confusing to me. Were subjects asked to close their eyes and move forearm by 'feel'?
Response 1: Thanks for the reviewer’s question. We have revised the description for proprioception and attention conditions in the experimental section (lines 226-240). We hope that the key issues of these studies can be explained. As follows: Because the proprioception is the feeling that the moving organs such as muscles, tendons, and joints appear on their own during exercise or at rest. For example, a person can sense the position of various parts of the body when closing his eyes. Therefore, our experiment required that the subject feeled the movement angle of the elbow joint with subjective feeling under the proprioceptive condition, and there was no visual feedback. The subjects were asked to close their eyes slightly and focus on the motion of the elbow joint in heart. The experimental setup of the internal and external attention conditions required visual feedback significantly. In the external focus of attention experiment, the subjects were asked to focus on a display screen (1280 × 800 px) with a front distance of 1 m. The sEMG signal and joint angle waveforms were displayed on the screen and they were the effect of motion. In the internal focus of attention experiment, the subjects were asked to visually notice their elbow joint motions. According to different psychological factors and motion directions, data from 6 experiments were collected, as shown in Table 1. The three psychological factors included external visual focus of attention (EVA), internal visual focus of attention (IVA), and proprioceptive no visual feedback (NVF). The two motion directions include a horizontal direction (HD) and a vertical direction (VD).
Point 2: Assuming I have understood this portion, it is quite surprising that data showed NVF (eyes closed, move by feel) yielded the best data (lowest overall error (Fig6) and low local error (figs4, 5).
Response 2: Thanks for the reviewer’s questions. The accuracy of elbow joint angle recognition under proprioceptive (NVF) conditions is higher than that of visual feedback (EVA and IVA). The wider force fluctuations in the presence of a continuous visual feedback were explained by continuous sensorial information, which possibly induced the subject to a greater number of adjustments and corrections of the force output. This also leads to a reduction in the degree of automatic movement.
Point 3: If I am understanding the experiment and data correctly, Discussion lines 302-304 are not supported by data. Accuracy is not improved by external visual data. In both cases, accuracy was improved by closing the eyes.
Response 3: Thanks for the reviewer’s suggestion. From the results of the angle recognition (RMSE value), it can be seen that the recognition accuracy under the NVF condition is the highest, followed by EVA, and finally IVA. In Lines 302-304, we compare EVA and IVA, but this expression is easily misunderstood. So in the revision, the NVF condition was first discussed, and we revised the sentences in Lines 348-350.
Point 4: Text font,and size varies throughout the document. Text in Figs 4, 5 are so small as to be illegible. Returning to the electronic copy and zooming in, the text is so pixelated as to be illegible. It is difficult to review a paper when the main results figures are illegible.
Response 4: Thanks for the reviewer’s suggestion. We have carefully revised Figures 4 and 5 for better reading.
Point 5: Figure 3 is quite large (could easily shrink to 1/2 the height but text is too small even at this size.
Response 5: Thanks for the reviewer’s suggestion. We have carefully revised Figures 3 for better reading.
Point 6: Figures4, 5 need to be completely redone. This is the heart of the paper, and it's difficult to tell what's going on. Why is green shaded in figure 5? Isn't that deviation from nominal? Then isn't distance from the red line the key aspect? Is the redline on zero degrees? can't read it.
Response 6: Thanks for the reviewer’s suggestion. Figures 4 and 5 have been redone. In Figure 5 we calculated the error of all estimated angles to the actual angle during the 10 second motion sampling time. The curves and shadows are intended to indicate which time period errors are greater in the time series of the entire motion process. By considering it again, we think that the choice of threshold (5°) must to be carefully justified. Because the error range lacks sufficient proof and significance. So we revised the reference to Burdea et al., and the relevant comments were also removed from the revision.
Point 7: Figure 7. it was never made clear the importance of Coherence. Probably eliminate it. If it remains, chop the top half. there is no data there.
Response 7: Thanks for the reviewer’s suggestion. Coherence is studied to explore whether there is synergy or interference between the brain and muscles in joint motion. It has been indicated in the revision (lines 166-167). Under the proprioceptive condition, there was almost no coherence between the two signals, and no crosstalk occurs between them. In combination with recognition accuracy, the automatic characteristics of motion and the flexibility of motion were not limited under the proprioceptive condition. Figure 7 has been revised.
Point 8: Figure 6: the only data is the six values. that could easily be in table or even text. The bar chart is large and conveys little information.
Response 8: Thanks for the reviewer’s suggestion. Figure 6 has been revised in the form of a table.
Point 9: It is difficult to comment on discussion and conclusions as the method was not clear. If my interpretation above is correct, then discussion is inaccurate. I hope I am simply misunderstanding lines 227-238.
Response 9: Thanks for the reviewer’s suggestion. We have carefully revised the description for proprioception and attention conditions. The discussion section has also been revised. The paper studied the accuracy of angle prediction can be affected by some psychological factors. The basis is that the elbow joint angle can be estimated by the sEMG signal during continuous motion using a classification or regression algorithm. For the elbow joint motion control based on the sEMG signal, at present, the classification recognition error using the motion type or the motion direction is relatively low. However, the features of the sEMG signal are converted to joint angle values through the regression method and then directly applied to the exoskeleton actuator, which is more able to ensure continuous matching of human-machine motion. This also effectively compensates for the inflexibility of classification identification. Although there is still a large error in angular regression, we have found that these errors can be appropriately reduced under some psychological conditions. The results of this paper can be used as a basis for more optimal sEMG control.
Point 10: I assume formatting was left for a future submission, but better figures would be very welcome. Figures 4 and 5 (the main data) were un-readable.
Response 10: Thanks for the reviewer’s suggestion. We have carefully revised Figures 4 and 5 for better reading.
Author Response File: Author Response.docx
Round 2
Reviewer 1 Report
Authors have properly addressed all my major concerns about the proposed manuscript. However, in my opinion the response to my first comment about the presentation of only one classification method should be inserted in the manuscript in order to make the reader aware of this partial limitation of the present study.
Author Response
Manuscript ID: applsci-436055
Title: " Effects of Proprioception and Visual Focus of Attention on Surface Electromyography Proportional Control "
Correspondence Author: Wang Weixing
Dear Reviewer:
Thank you very much for a rapid processing of our manuscript exclusively submitted for publication in Applied sciences. We thank the reviewers for their constructive criticism that has helped us to improve the manuscript. The summary of the changes and responses to your comments are listed below. We sincerely hope this manuscript will be finally acceptable to be published on  Applied sciences.
Thank you very much for all your help and looking forward to hearing from you soon.
Best regards
Sincerely yours
Wang Weixing
Corresponding Author
Response to Reviewer 1 Comments
Point 1: Authors have properly addressed all my major concerns about the proposed manuscript. However, in my opinion the response to my first comment about the presentation of only one classification method should be inserted in the manuscript in order to make the reader aware of this partial limitation of the present study.
Response 1:
Thanks for the reviewer’s suggestion very much. We have already explained in the revised section 2.3 that only one classification method has been used (lines 128-132). As follow: In this paper, BP neural network (BPNN) was chosen for joint angle regression. Prior to this, Support vector machines (SVM) and Random forest algorithms (RF) were also used to compare the accuracy of regression recognition. The results showed that the BPNN algorithm is relatively good. Because our research focuses on the effects of several psychological conditions, only the calculation results of BP neural network are placed in the paper.
In addition, the relevant description is also inserted in Section 6 (lines 383-385, 392-396). As follow: Although there is still a large error in the accuracy of the regression of the elbow joint angle, we believe that pattern recognition based sEMG proportional control will likely be implemented and become more stable in the near future. … Secondly, through the verification of this paper, it is found that the use of traditional pattern recognition algorithms including BPNN, SVM and RF has certain limitations in the accuracy of sEMG-based angle regression. In order to reduce the recognition error, we hope to study the sEMG array data from more channels of biosignal acquisition devices, and deep learning algorithms are used for processing.
Author Response File: Author Response.docx
Reviewer 2 Report
with current edits, article looks good.
Author Response
Manuscript ID: applsci-436055
Title: " Effects of Proprioception and Visual Focus of Attention on Surface Electromyography Proportional Control "
Correspondence Author: Wang Weixing
Dear Reviewers:
Thank you very much for a rapid processing of our manuscript exclusively submitted for publication in Applied Sciences. We thank the reviewers for their constructive criticism that has helped us to improve the manuscript. We sincerely hope this manuscript will be finally acceptable to be published on  Applied Sciences.
Thank you very much for all your help and looking forward to hearing from you soon.
Best regards
Sincerely yours
Wang Weixing
Corresponding Author
Response to Reviewer 2 Comments
Point 1: with current edits, article looks good.
Response 1: We appreciate for Reviewers’ warm work earnestly. Once again, thank you very much for your comments and suggestions.
Author Response File: Author Response.docx
