Deep Neural Network for Lung Image Segmentation on Chest X-ray

Round 1

Reviewer 1 Report

This paper introduces Res-UNet architecture for lung image segmentation. The overall manuscript is poorly prepared. I recommend to reject this manuscript from the aspects of manuscript quality, model, optimization, and data:

• The overall quality of writing should be improved, since countless language and presentation issues are found (even in abstract):
  • Terminologies used across the whole manuscript should be consistent. E.g., ‘Res-UNet++’ is used in line 18, while ‘ResUnet++’ is used in line 20.
  • Redundant sentences are found, e.g., ‘This paper’s novelty …’ in line18 and ‘The novelty of this research paper …’ in line 24.
  • Keywords are not appropriately chosen. E.g., if ‘CNN’ is included as a keyword, then ‘Convolutional’ is not necessary. Besides, ‘model’ is not suitable to be a keyword.
  • Paragraphs and titles are not appropriately indented.
• The novelty of the proposed architecture is weak. First of all, Unet itself uses residual links to pass fine-grained information from encoder to decoder, thus it can be regarded as a special case of ResNet. In this view, a name like ‘ResUnet’ is not appropriate. Secondly, the motivation of using classic residual block in a Unet should be discussed more clearly, because integrating a bunch of fancy structures sometimes leads to bad results.
• In deep learning study, if one claims that one model consistently outperforms another, one should thoroughly perform hyper-parameters searching for each model. Otherwise, the results will not be convincing. If such experiments are not carried out, at least provide the codes for reproducing the results given in the paper.
• The authors claim in the abstract that they will introduce how to accurately identify diseased regions in lung images. However, according to the rest part of this paper, the task seems to be simply segmenting out lung area from each lung image. </aside>

Author Response

Reviewer 1:

Comment 1 – The overall quality of writing should be improved, since countless language and presentation issues are found (even in abstract): Terminologies used across the whole manuscript should be consistent. E.g., ‘Res-UNet++’ is used in line 18, while ‘ResUnet++’ is used in line 20. Redundant sentences are found, e.g., ‘This paper’s novelty ...’ in line18 and ‘The novelty of this research paper ...’ in line 24.

Response- Thanks for the comment and points of improvement. We have used Grammarly professional and worked on the consistent representation of the terminologies used in the paper.

Action-We have worked carefully on the points suggested in the comment 1 and used consistent terminologies in the paper. Grammarly professional and proofreading is also used to avoid minor mistakes in the paper.

 

Comment 2- Keywords are not appropriately chosen. E.g., if ‘CNN’ is included as a keyword, then ‘Convolutional’ is not necessary. Besides, ‘model’ is not suitable to be a keyword.

Response- Thanks for the comment.

Action – We have carefully chosen new keywords such as nCoV, COVID-19, CNN.

 

Comment 3- Paragraphs and titles are not appropriately indented.

Response- Thanks for the comment. We have improved the formatting carefully and made it representable.

Action - We have improved the formatting carefully and made it representable. The new title is now ‘Deep Neural Network for Lung image segmentation on Chest X-Ray’

 

 

Reviewer 2 Report

I suggested this paper to be accepted because its novelty. 

Author Response

Comment-I suggested this paper to be accepted because its novelty. 

 

Response-Thank you so much for your time and acceptance.

Reviewer 3 Report

I think this research is good. However, there are some problems.

1. subject in the sentence 

In the paper, if you describe something that you have done, you should better use “we” instead of using "they" and "authors". For example, in lines 107 and 134. Please address this problem in the whole manuscript. Following, I listed some examples that confused me.

In line 240, you said"they", do you mean "yourselves"?

In line 169, you said, "the author proposed". I wonder if the network is proposed by other authors in the previous paper (existing method) or you. If it is proposed by you, you should better say "we proposed". 

Please address this kind of problem throughout the whole manuscript.

2. discussion part

The discussion part needs to be expanded. Even though this manuscript has a "results and analysis" part, I think you can discuss your experiments and results in the discussion part instead of only a few sentences.

3. introduction part

The literature review part can be expanded.

4.conclusion part

In line 294, you said "various datasets", but you introduced one dataset in section 2a. I am not sure the number of the dataset you used in the research, please  check.

Author Response

Reviewer 3:

Comment 1 –

In the paper, if you describe something that you have done, you should better use “we” instead of using "they" and "authors". For example, in lines 107 and 134. Please address this problem in the whole manuscript. Following, I listed some examples that confused me. In line 240, you said"they", do you mean "yourselves"? In line 169, you said, "the author proposed". I wonder if the network is proposed by other authors in the previous paper (existing method) or you. If it is proposed by you, you should better say "we proposed". Please address this kind of problem throughout the whole manuscript.

Response- Thanks for the comment. We have improved the comment and have made changes wherever applicable.

 

Action- We have made changes in full manuscript according to the comment.

 

Comment 2 – The discussion part needs to be expanded. Even though this manuscript has a "results and analysis" part, I think you can discuss your experiments and results in the discussion part instead of only a few sentences.

Response- Thanks for the comment. We have expanded the discussion part and done changes as suggested.

Action – We have elaborated the Discussion part in section…on page no…

 

Comment 3 – Introduction Part: “The literature review part can be expanded.”

Response- Thanks for the comment. We have expanded the literature review section as suggested in the comment.

Action- We have elaborated the literature review section …on page no…

Comment 4 – Conclusion part “In line 294, you said "various datasets", but you introduced one dataset in section 2a. I am not sure the number of the dataset you used in the research, please check.”

Response- Thanks for the comment. We have improved the comment and have made changes wherever applicable.

Action – According to a comprehensive assessment utilizing chest X-Ray datasets, the ResUNet++ design outperforms the state-of-the-art UNet and ResUNet architectures in delivering semantically correct predictions.

 

Round 2

Reviewer 3 Report

When you describe your work in this manuscript, you do not need to use "we have propose……in this paper". You can say "we propose……in this paper". Please use the present simple tense to describe your work in the paper.

Author Response

When you describe your work in this manuscript, you do not need to use "we have propose……in this paper". You can say "we propose……in this paper". Please use the present simple tense to describe your work in the paper.

Response: Thanks for the correction. We have made necessary changes in the draft and uploading the updated manuscript.  

 

Author Response File: Author Response.docx