A Lightweight CNN and Class Weight Balancing on Chest X-ray Images for COVID-19 Detection

Round 1

Reviewer 1 Report (Previous Reviewer 1)

After closely reviewing the entire article, I noticed various flaws. Before publishing, authors should think about the following points:

1-In terms of hierarchy, missing, gap, background how and what, the introduction part must be updated.

2-The introduction section also lacks sufficient citations. The authors are suggested to use these six sources and cite them when discussing topics that go beyond the scope of this paper.

https://www.mdpi.com/2079-9292/10/16/1996

https://link.springer.com/article/10.1007/s00521-022-07424-w

https://www.sciencedirect.com/science/article/pii/S0010482522002530

https://www.mdpi.com/2079-9292/11/19/3113

https://www.sciencedirect.com/science/article/pii/S0010482521009355

https://www.mdpi.com/2079-9292/11/17/2682

3-"At the time 16 September 2022, 607 745 726 people 18 have tested positive, 6 498 747 people have died, with 12 540 061 501 Vaccine doses [1] 19 https://www.who.int/. It has signs, and symptoms like Pneumonia [2] however is extra 20 fatal than that. Early diagnosis and adequate safety measures are required to limit the 21 transmissibility of COVID-19 infection [3,4]." Authors stated. Links and controversial statements should be removed from the paper.

4-The entire paper must be rechecked in terms of English.

5-Sort works in chronological order first in the related work section, then add a related work table. Finally, in the related work table, include your work to demonstrate its own novelty.

6-No section should be left blank. For example, the authors suddenly switched from sections 3 to 3. 1. Fill up section 2 with appropriate sentences.

7-If formulas are borrowed from other works, they must be cited.

8-In addition, the method's disadvantages must be stated in the conclusion section.

Author Response

Original Manuscript ID: electronics-2032828

Original Article Title:  

A lightweight CNN and class weight balancing on Chest X-ray images for Covid-19 detection

Dear Editor and reviewers

We are thankful for the time and effort spent by the referee reviewing our manuscript. I appreciate your kindness that is still giving us a chance to revise our manuscript. All the suggested changes have been duly incorporated in the revised version of the paper. All new changes that were added, based on the reviewers’ feedback, are displayed in different colours. The English language has been revised as requested. Below are all the comments we received, in addition to our actions.

Author Response File: Author Response.pdf

Reviewer 2 Report (Previous Reviewer 2)

The revision has addressed all my concerns.

Author Response

Original Manuscript ID: electronics-2032828

Original Article Title:  

A lightweight CNN and class weight balancing on Chest X-ray images for Covid-19 detection

Dear Editor and reviewers

We are thankful for the time and effort spent by the referee reviewing our manuscript.

I appreciate your kindness that is still giving us a chance to revise our manuscript. All the suggested changes have been duly incorporated in the revised version of the paper. 

All new changes that were added, based on the reviewers’ feedback, are displayed in different colours. The English language has been revised as requested. Below are all the comments we received, in addition to our actions.

Author Response File: Author Response.pdf

Reviewer 3 Report (New Reviewer)

The following comments must be carefully revised.

(1) The following work (e.g., regularization and generalization of deep learning, medical applications, etc) should be cited and discussed in Introduction, including “Improvement of generalization ability of deep CNN via implicit regularization in two-stage training process,” IEEE Access, vol. 6, pp. 15844-15869, 2018. “Faster Mean-shift: GPU-accelerated clustering for cosine embedding-based cell segmentation and tracking.” Medical Image Analysis 71 (2021): 102048. “VoxelEmbed: 3D instance segmentation and tracking with voxel embedding based deep learning,” International Workshop on Machine Learning in Medical Imaging. Springer, Cham, 2021. “Pseudo RGB-face recognition,” IEEE Sensors Journal, 2022, doi: 10.1109/JSEN.2022.3197235.

(2) The specific model of deep learning (e.g. the number of convolution kernels, step size, etc) and the reason for designing this architecture should be further introduced.

(3) The optimization curve of the objective function of the deep learning model with the iterative process of various settings should be all presented.

(4) The experimental setup of the proposed deep learning model and comparison methods should be reported in detail, such as batch size.

(5) The structural complexity and reasoning speed of the model are encouraged to display.

(6) More technical details should be added in the second part to help understand the method used.

(7) The results of TP value of COVID in Fig. 10 should be further explained.

(8) Some figures are blurry and cannot be seen clearly.

Author Response

Original Manuscript ID: electronics-2032828

Original Article Title:  

A lightweight CNN and class weight balancing on Chest X-ray images for Covid-19 detection

Dear Editor and reviewers

We are thankful for the time and effort spent by the referee reviewing our manuscript.

I appreciate your kindness that is still giving us a chance to revise our manuscript. All the suggested changes have been duly incorporated in the revised version of the paper. 

All new changes that were added, based on the reviewers’ feedback, are displayed in different colours. The English language has been revised as requested. Below are all the comments we received, in addition to our actions.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report (Previous Reviewer 1)

It can b accepted now.

Reviewer 3 Report (New Reviewer)

All comments are carefully revised and the quality of this paper has been significantly improved. Therefore, this paper can be accepted for publication.

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 authors provided a simple CNN with CWB for classifying chest X-ray images. This article tries to change CNN's capability to display chest X-ray images in addition to the fact that there is a class disparity.  Here are my observations:

1. The paper has a weak structure. Overall, nearly the entire article has to be revised. Paragraphs in introductions, for instance, are uneven.

2-The introduction section contains a number of problems. Missing are the hierarchy, gap, purpose, and method. 

3-The related work section is not at all meaningful. It lacks any supporting scientific evidence. Rewrite this part to include a related work table and add your work to highlight your originality.

4-If formulas borrowed from other works must be properly referenced.

5-The suggested approach must be explained in detail by the authors.

6-Experiments and findings are in their infancy. The results must be explained by the authors. additional tables and figures should be included. What is the drawback of your work? It must be brought up.

7-The figures' quality is totally inappropriate. for example, Figs. 1, 2, 3, 10, and so on. recreate them.

8-The authors might discuss the significance of deep learning techniques and the wider healthcare establishment in the introduction section. In the introduction section, the authors encouraged the utilization of these important sources.

https://www.mdpi.com/2076-3417/11/8/3414

https://www.sciencedirect.com/science/article/pii/S0010482522002530

https://link.springer.com/article/10.1007/s00521-022-07424-w

https://link.springer.com/article/10.1186/s40537-020-00392-9%23auth-Taghi_M_-Khoshgoftaar

https://www.sciencedirect.com/science/article/pii/S2666990021000240

Author Response

Original Manuscript ID:   electronics-1922735

Original Article Title:  

A lightweight CNN and class weight balancing on Chest X-ray images for Covid-19 detection

Dear Editor:

We are thankful for the time and efforts spent by the referee reviewing our manuscript. Thanks a lot for your patience with our imperfect manuscript and your great efforts in polishing our manuscript.

I appreciate your kindness that you still give us a chance to revise our manuscript. All the suggested changes have been duly incorporated in the revised version of the paper.

All new changes that were added, based on the reviewers’ feedback, are displayed in different colours.

The English language has been revised; we hope the quality of the English language is satisfactory. Below are all the comments we received, in addition to our actions.

Author Response File: Author Response.pdf

Reviewer 2 Report

The contributions of the paper are unclear. I would suggest to add a new paragraph in the introduction section to clarify the main contributions of the article. 

In the related work section, many recent works about chest x-ray based thoracic disease agnostic are missed, such as, Contrast-Attentive Thoracic Disease Recognition with Dual-Weighting Graph Reasoning and Many-to-One Distribution Learning and K-Nearest Neighbor Smoothing for Thoracic Disease Identification.

Further, experimental comparisons against these recent methods should be provided.

There are some large-scale chest x-ray datasets like chestx-ray14 and chestxpert, however, only two small-scale datasets are used in the paper. More discussions are needed to clarify the motivation here.

The caption of Table 8 should be re-edited.

Author Response

Original Manuscript ID: electronics-1922735

Original Article Title:  

A lightweight CNN and class weight balancing on Chest X-ray images for Covid-19 detection

Dear Editor:

We are thankful for the time and efforts spent by the referee reviewing our manuscript. Thanks a lot for your patience with our imperfect manuscript and your great efforts in polishing our manuscript. 

I appreciate your kindness that you still give us a chance to revise our manuscript. All the suggested changes have been duly incorporated in the revised version of the paper. 

All new changes that were added, based on the reviewers’ feedback, are displayed in different colours. 

The English language has been revised; we hope the quality of the English language is satisfactory.

Below are all the comments we received, in addition to our actions.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Unfortunately, I can't perceive any positive modifications in the article. To understand how to write a scientific article, the authors must read additional publications.

Reviewer 2 Report

The revision has addressed all my concerns. I recommend to accept the paper.