Pneumonia Recognition by Deep Learning: A Comparative Investigation

Round 1

Reviewer 1 Report

 My comments in the first round of review have all been well answered, and I think the paper can be accepted.

Author Response

Thank you so much again for your constructive and insightful comments!

Author Response File: Author Response.pdf

Reviewer 2 Report

The authors have addressed my concerns.

Author Response

Thank you so much again for your constructive and insightful comments!

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript under the title "Pneumonia Recognition by Deep Learning: A Comparative Investigation" seems a good work. However, novelty of the highlighted paper must be improvement. In revised version of manuscript , the few previous highlighted concerns are not properly incorporated. Authors must consider the below concerns.

1. In introduction section, the figure organogram is still missing. Moreover,  the contribution highlighted in the end of introduction must be improved.
2. Major concern:  workflow in figure 1 is still confusing especially the training and testing of deep learning models 
3. Major concern:  table 2 of time complexity is vague one. The calculation of time complexity must be proved with mathematical modeling
4. Major concern, justification of time complexity is missing , authors must be improved 
5. The BIG O notation calculation must be discussed in separate section and must be compared with state of the art studies.
6. Section 2.2.1 and section 2.2.2 must be improved , authors must be improved.
7.  There are various grammatical errors exist in whole paper, authors must be improved   

Author Response

Dear Reviewer,

We would like to submit our paper entitled “Pneumonia Recognition by Deep Learning: A Comparative Investigation” for your consideration for publication in the journal Applied Sciences.

The manuscript is a resubmission after revision. The old Manuscript ID is applsci-1612257.

We have made a point-by-point response to the reviewers’ comments and suggestions, including a detailed description of any requested or suggested revisions.

We have also carefully checked and corrected the writing format and errors to make our revised manuscript conform to the journal style.

All the modifications and explanations in this revised version are listed in detail in the following “Responses to Reviewer's Comments”.

We would deeply appreciate your consideration and reviewers’ helpful comments and suggestions.

Yours Sincerely,

Yuting Yang, Gang Mei*

School of Engineering and Technology, China University of Geosciences (Beijing)

Email: [email protected] (G. Mei)

 

Author Response File: Author Response.pdf

Reviewer 4 Report

This paper compares 5 different DL models for pneumonia detection. This paper has potential as it helps identify the best DL model for such detection; however it requires certain issues to be fixed before acceptance.

1. The motivation in the introduction could be improved further. For example, why the comparison is required, and are there any comparative study papers?
2. The evaluation results require precision, recall, f1-score in addition to accuracy for the comparison purpose. Please use them for the comparison.
3. The paper is missing recent SOTA papers strongly related to COVID-19 pneumonia detection using the DL method. I encourage authors to put and explain those three recent papers in this paper. *Attention-based VGG-16 model for COVID-19 chest X-ray image classification | SpringerLink, *New bag of deep visual words based features to classify chest x-ray images for COVID-19 diagnosis | SpringerLink, *Fusion of multi-scale bag of deep visual words features of chest X-ray images to detect COVID-19 infection | Scientific Reports (nature.com)

 These recent three papers still employ the VGG-16 DL model for pneumonia detection. Please explain them in the paper.

 

Author Response

Dear Reviewer,

We would like to submit our paper entitled “Pneumonia Recognition by Deep Learning: A Comparative Investigation” for your consideration for publication in the journal Applied Sciences.

The manuscript is a resubmission after revision. The old Manuscript ID is applsci-1612257.

We have made a point-by-point response to the reviewers’ comments and suggestions, including a detailed description of any requested or suggested revisions.

We have also carefully checked and corrected the writing format and errors to make our revised manuscript conform to the journal style.

All the modifications and explanations in this revised version are listed in detail in the following “Responses to Reviewer's Comments”.

We would deeply appreciate your consideration and reviewers’ helpful comments and suggestions.

Yours Sincerely,

Yuting Yang, Gang Mei*

School of Engineering and Technology, China University of Geosciences (Beijing)

Email: [email protected] (G. Mei)

 

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report

I would like to thank the authors for their hard work to improve the quality of the manuscript. The reviewer to inclined to accept the manuscript.

Just a minor comment, the doi of reference number 25 is redirecting to the correction file, not the original paper. 

The complete bib entry of this paper is:

@article{sitaula2021fusion, title={Fusion of multi-scale bag of deep visual words features of chest X-ray images to detect COVID-19 infection}, author={Sitaula, Chiranjibi and Shahi, Tej Bahadur and Aryal, Sunil and Marzbanrad, Faezeh}, journal={Scientific reports}, volume={11}, number={1}, pages={1--12}, year={2021}, publisher={Nature Publishing Group} }

 

Author Response

Thank you so much for your valuable comments and approval!
We have revised the reference number 25.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The manuscript under the title " Pneumonia Recognition by Deep Learning: A Comparative Investigation" is a average work. There are few major concerns that listed below. 1. The structure of abstract is quite confusing in sense of authors contribution 2. Why the need of such comparative Investigation ?, if there are some limitations in existing methods then authors must add the limitation section and properly discuss the challenges along limitations of existing methods. 3. The figure 1 of proposed method is not clear , is step 2 is automated or not ?? 4. Section 2.3.2 GPU loading is not properly discussed, authors need to compared execution time of deep learning models in the form of BIG O notations 5. Authors thoroughly discuss the chest X-ray image datasets and design one more table that shows the features of different chest X-ray image datasets 6. Additionally, authors need to add the more related studies of 2021 and also cite few studies of 2022

Author Response

Dear Reviewer,

We would like to submit our paper entitled “Pneumonia Recognition by Deep Learning: A Comparative Investigation” for your consideration for publication in the journal Applied Sciences.

We have made a point-by-point response to the reviewers’ comments and suggestions, including a detailed description of any requested or suggested revisions.

We have also carefully checked and corrected the writing format and errors to make our revised manuscript conform to the journal style.

All the modifications and explanations in this revised version are listed in detail in the following “Responses to Reviewer's Comments”.

We would deeply appreciate your consideration and reviewers’ helpful comments and suggestions.

Yours Sincerely,

Yuting Yang, Gang Mei*

School of Engineering and Technology, China University of Geosciences (Beijing)

Email: [email protected] (G. Mei)

 

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper evaluates different deep learning models on pneumonia datasets to decide the optimal model.

The paper does not have any novel contribution. The paper neither provides a new dataset nor a new model. Furthermore, the dataset used is small as pointed out by the authors and that could also impact the results.

Considering the contribution of this work. the authors have only run different models on existing datasets and shown the results. The conclusions provided are also not novel. The impact of GPU loading and data augmentation has been well studied in the literature.

It is important to point out the reason behind inspecting the said set of models to address the problem. Why not use other models and only these five?

The authors have listed some existing approaches for pneumonia detection, but there is no comparative study. The authors should either provide a comparative study or discuss why it is not possible.

Author Response

Dear Reviewer,

We would like to submit our paper entitled “Pneumonia Recognition by Deep Learning: A Comparative Investigation” for your consideration for publication in the journal Applied Sciences.

We have made a point-by-point response to the reviewers’ comments and suggestions, including a detailed description of any requested or suggested revisions.

We have also carefully checked and corrected the writing format and errors to make our revised manuscript conform to the journal style.

All the modifications and explanations in this revised version are listed in detail in the following “Responses to Reviewer's Comments”.

We would deeply appreciate your consideration and reviewers’ helpful comments and suggestions.

Yours Sincerely,

Yuting Yang, Gang Mei*

School of Engineering and Technology, China University of Geosciences (Beijing)

Email: [email protected] (G. Mei)

 

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors compared five deep learning pneumonia recognition models to filter out the optimal model for pneumonia recognition in various cases, thus making the practical application and model selection of deep learning models for pneumonia recognition more convenient. It is a well-structured paper with interesting results. However, it requires further improvements before publication.

1.The abstract should be narrow down on the problem and highlight the need of the proposed work with experimental results.

2.In the introduction, the novelty and the contributions of your works are given.

3.The literature review is poor in this paper. You must review all significant similar works that have been done. I hope that the authors can add some new references in order to improve the reviews and the connection with the literatures. For example, 10.1109/TCYB.2020.3033005; 10.1016/j.ins.2021.11.052; 10.1109/ACCESS.2021.3108972 and so on.

4. The font is too large in Figure 1. Please change it to a suitable font.
5. The method/approach in the context of the proposed work should be written in detail.
6. The values of parameters could be a complicated problem itself, how the authors give the values of parameters in the used methods.
7. The authors need to interpret the meanings of the variables.
8. Proofread the paper carefully to improve it grammatically. For example, “are compared in various cases to filter out the optimal model for pneumonia recognition for application in actual pneumonia recognition”, ….
9. Where does the data come from?

Author Response

Dear Reviewer,

We would like to submit our paper entitled “Pneumonia Recognition by Deep Learning: A Comparative Investigation” for your consideration for publication in the journal Applied Sciences.

We have made a point-by-point response to the reviewers’ comments and suggestions, including a detailed description of any requested or suggested revisions.

We have also carefully checked and corrected the writing format and errors to make our revised manuscript conform to the journal style.

All the modifications and explanations in this revised version are listed in detail in the following “Responses to Reviewer's Comments”.

We would deeply appreciate your consideration and reviewers’ helpful comments and suggestions.

Yours Sincerely,

Yuting Yang, Gang Mei*

School of Engineering and Technology, China University of Geosciences (Beijing)

Email: [email protected] (G. Mei)

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The revised manuscript under the title " Pneumonia Recognition by Deep Learning: A Comparative Investigation" seems improved as compared to previous version. However, there are still much improvement require for the publication.  Few serious concerns highlighted in below 

1. The actual contribution of this study is missing in this study, authors must add  the list of contributions of this study in the end of introduction section.
2.  In last of introduction section, authors must highlight the organogram diagram for better representation of rest sections of articles 
3.  The  detail  literature  review section still required in revised of manuscript. Furthermore, authors must incorporated one table that clearly highly the contributions of the previous studies as well limitations of previous studies .
4. Major concern: authors must  explain the workflow figure 1 in algorithmic manner.
5.  Moreover,  the detail discussion on figure 1 still require.
6. Authors clearly highlight the evaluation parameters .
7.  In section 2.2.1,  data collection and pre- processing phase not explain in right way, authors requires to revise the whole section.
8. In section 2.2.2,  data augmentation is not properly discuss in revised version 
9. Table 1 is not represented in right way, the complete method of computation complexity in terms of BIG O is required at stage.
10.  The state of the art comparison of BIG O is highlighted in terms of  models epoch values

Reviewer 2 Report

1. The comparison part of the paper is yet not clear. The existing works the authors have cited do have limitations but are designed for pneumonia detection. A comparison with these models is important to make the conclusions that are theorized.
2. Models like VGG and InceptionNet work similarly to the models discussed in this work and are more complex. They also provide better results in specific scenarios. Why would those not be included in the study?

Reviewer 3 Report

I have appreciated the deep revision of the contents and the present form of this manuscript. All my previous concerns have been accurately addressed. I think that this paper can be accepted.