Machine Learning Styles for Diabetic Retinopathy Detection: A Review and Bibliometric Analysis

Round 1

Reviewer 1 Report

Good technical contribution to the topic is made by the authors. Well done. A reasonable collection of documents are referred for this systematic literature review and analysis performed makes a significant impact to this study.

The paper requires a major review of the language, grammar and punctuation throughout the paper.

The entire paper requires a careful proof reading, for example, section 4.2 heading has an un necessary"." in the beginning of the title as shown here "4.2. . Steps in Diabetic Retinopathy Detection"

Author Response

The paper requires a major review of the language, grammar and punctuation throughout the paper.

Yes, it has been done

The entire paper requires a careful proof reading, for example, section 4.2 heading has an un necessary"." in the beginning of the title as shown here "4.2. . Steps in Diabetic Retinopathy Detection"

Yes, removed

Reviewer 2 Report

It is a good review paper that presents a clear trend in Diabetic Retinopathy Detection with machine learning methods. It could be accepted if the following issues can be addressed:

1.      In lines 35-37, please add the reference for the statement or the data source.

2.      In line 105, figure 6 did not indicate any date. Please specify the date or modify the figure.

3.   Figure 13, it is hard to match the size of those blue dots in the legend with them in the figure. And please use a different color to separate the journal articles and conference papers.

4.      Please add the figure legend like figure 13 for figure 14.

5.      In line 197, please modify the statement this way: “… forming a single cluster indicated by green area in the figure”. And add a similar description for the second cluster and the third cluster.

6.      In figure 16, how these clusters get formed. Is it the method indicated at the top of the figure which is MCA? Please explain this method or add a reference for it.

7.      The page index disappears after page 10 and it starts from 2 on page 16, please fix this issue.

8.      In line 208, too many spaces between “3.10” and “Thematic Evolution Analysis”.

9.      In figure 18, the y-axis titles are blocked, please adjust these figures.

10.   In lines 260-263, it seems only e-Ophtha EX is discussed, how about e-Ophtha MA listed in the table3? What is the difference between these two?

11.   In line 277, please add references for datasets DRIVE and STARE.

12.   In line 404, the font style should be adjusted

13.   In section 4.5, it is better to analyze table 11 like which paper or method performs best when the lesion type is EX or which paper get best performance with which dataset.

Author Response

In lines 35-37, please add the reference for the statement or the data source.

2. 2.      In line 105, figure 6 did not indicate any date. Please specify the date or modify the figure.
3.  Figure 13, it is hard to match the size of those blue dots in the legend with them in the figure. And please use a different color to separate the journal articles and conference papers.
4. 4.      Please add the figure legend like figure 13 for figure 14.
5. 5.      In line 197, please modify the statement this way: “… forming a single cluster indicated by green area in the figure”. And add a similar description for the second cluster and the third cluster.
6. 6.      In figure 16, how these clusters get formed. Is it the method indicated at the top of the figure which is MCA? Please explain this method or add a reference for it.
7. The page index disappears after page 10 and it starts from 2 on page 16, please fix this issue.
8. 8.      In line 208, too many spaces between “3.10” and “Thematic Evolution Analysis”.
9. In figure 18, the y-axis titles are blocked, please adjust these figures.
10. In lines 260-263, it seems only e-Ophtha EX is discussed, how about e-Ophtha MA listed in the table3? What is the difference between these two?
11. In line 277, please add references for datasets DRIVE and STARE.
12. In line 404, the font style should be adjusted
13. 13.   In section 4.5, it is better to analyze table 11 like which paper or method performs best when the lesion type is EX or which paper get best performance with which dataset.

Yes all the above points are rectified

Reviewer 3 Report

This review focused on machine learning styles for diabetic retinopathy detection. The authors did a lot of work, and the review is well-written. I have some minor points, please see below.

1. The number of Figures was too many. Please combine them properly.

2. For Figures 1 to 3, please add the source of data.

3. The authors should emphasize the advantages and significance of new technologies such as machine learning in the section of the Introduction.

4. It would be great if the authors could add a section discussing the limitation and accuracy of different machine learning styles for diabetic retinopathy detection

Author Response

1. The number of Figures was too many. Please combine them properly.
2. For Figures 1 to 3, please add the source of data.
3. The authors should emphasize the advantages and significance of new technologies such as machine learning in the section of the Introduction.

4. It would be great if the authors could add a section discussing the limitation and accuracy of different machine learning styles for diabetic retinopathy detection

Yes, done

Our focus is mainly on bibliometric analysis. To understand bibliometric analysis a qualitative study is done.

Reviewer 4 Report

The paper is focused on performing a review and bibliometric analysis on the most recently developed tools that use machine learning for the detection of diabetic retinopathy. The topic is interesting and of a high interest for the research community. Furthermore, the paper includes several graphics that are part of the bibliographic analysis tools in the research area, and help the reader to reach an overview on the research area.

Overall, we think that the main feature and merit of the paper is to providing an screenshot on such topic, instead of going deep into specific techniques to performing the detection of diabetic retinopathy.

Before its possible publication, we think that the paper should be improved in some directions:

-The introduction section should justify in a better way the development of this survey. Being a very hot research topic nowadays, recently several surveys has been detected in the same area [1-4]. Even though some of them are focused on "image processing", this viewpoint cannot be separated from the "machine learning" viewpoint, regarding at last, the detection of diabetic retinopathy works over images. What is added value of this survey? Why it is worthy to be published, beyond the previously published work [1-4]?

[1] Stolte, S., & Fang, R. (2020). A survey on medical image analysis in diabetic retinopathy. Medical image analysis, 64, 101742.

[2] Hasan, D. A., Zeebaree, S. R., Sadeeq, M. A., Shukur, H. M., Zebari, R. R., & Alkhayyat, A. H. (2021, April). Machine Learning-based Diabetic Retinopathy Early Detection and Classification Systems-A Survey. In 2021 1st Babylon International Conference on Information Technology and Science (BICITS) (pp. 16-21). IEEE.

[3] Bilal, A., Sun, G., & Mazhar, S. (2021). Survey on recent developments in automatic detection of diabetic retinopathy. Journal Français d'Ophtalmologie, 44(3), 420-440.

[4] Atwany, M. Z., Sahyoun, A. H., & Yaqub, M. (2022). Deep learning techniques for diabetic retinopathy classification: A survey. IEEE Access.

-Some tables and images in the manuscript should be improved, considering that the tables have been presented as images, making difficult the link of its content to the reference section.

-The discussion and conclusions section should be improved. It is necessary to point out what are the next steps in the research are, based on the findings detected in the survey. We also suggest to point out to recent trends in machine learning, such as explanations in e-health scenarios [5-6].

[5] Yera, R., Alzahrani, A. A., & Martínez, L. (2022). Exploring post-hoc agnostic models for explainable cooking recipe recommendations. Knowledge-Based Systems, 251, 109216.

[6] Shao, Y., Cheng, Y., Shah, R. U., Weir, C. R., Bray, B. E., & Zeng-Treitler, Q. (2021). Shedding light on the black box: explaining deep neural network prediction of clinical outcomes. Journal of medical systems, 45(1), 1-9.

We suggest major revision.

Author Response

Our focus is mainly on bibliometric analysis. To understand bibliometric analysis, a preliminary qualitative study is done.

Round 2

Reviewer 4 Report

We think that the presented study have a high level and illustrates the knowledge of the authors in relation to the machine learning-based diabetic retinopathy detection.

However, we consider that there are important issues that have not been considered by the authors yet, and that still apply for the current version of the manuscript:

-The introduction section should justify in a better way the development of this survey. Being a very hot research topic nowadays, recently several surveys has been detected in the same area [1-4]. Even though some of them are focused on "image processing", this viewpoint cannot be separated from the "machine learning" viewpoint, regarding at last, the detection of diabetic retinopathy works over images. What is added value of this survey? Why it is worthy to be published, beyond the previously published work [1-4]?

[1] Stolte, S., & Fang, R. (2020). A survey on medical image analysis in diabetic retinopathy. Medical image analysis, 64, 101742.

[2] Hasan, D. A., Zeebaree, S. R., Sadeeq, M. A., Shukur, H. M., Zebari, R. R., & Alkhayyat, A. H. (2021, April). Machine Learning-based Diabetic Retinopathy Early Detection and Classification Systems-A Survey. In 2021 1st Babylon International Conference on Information Technology and Science (BICITS) (pp. 16-21). IEEE.

[3] Bilal, A., Sun, G., & Mazhar, S. (2021). Survey on recent developments in automatic detection of diabetic retinopathy. Journal Français d'Ophtalmologie, 44(3), 420-440.

[4] Atwany, M. Z., Sahyoun, A. H., & Yaqub, M. (2022). Deep learning techniques for diabetic retinopathy classification: A survey. IEEE Access.

Author Response

-The introduction section should justify in a better way the development of this survey.

The development of this survey has been added in the introduction section.

What is added value of this survey? Why it is worthy to be published, beyond the previously published work [1-4]?

The added value of this survey is a thorough bibliometric analysis and machine learning categorization. The ML categorization that we have done is different from the mentioned studies. Our study has broad categorization under Six headings. And this comprehensive study will be helpful for researchers to understand in a better way.