VLSM-Net: A Fusion Architecture for CT Image Segmentation

Round 1

Reviewer 1 Report

The study proposes a new performance enhancement method for semantic segmentation in medical tomography images made in a very popular field.

 

1- Current competitors have not been adequately examined in the study. Tons of publications and codes are available at the following link: https://paperswithcode.com/dataset/luna16

2- Why weren't alternatives such as U-net preferred?

3- The v-net architecture proposed in the method section should be described in more detail.

4- It would be good to share semantic segmentation learning curve and IoU metric.

5- Benchmarking should be done with current competitors and presented in the discussion section.

6- Data and code should be shared with the reviewer and/or editor for verification. Blind review status should not be ignored.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

Reviewer’s Comments:

The manuscript “CT Image Segmentation Based on Fusion of V-Net and Level Set” is a very interesting work. In this work, region of interest (ROI) segmentation is a key step in computer-aided diagnosis (CAD). With the problems of blurred tissue edges and imprecise boundaries of ROI in medical images, it’s hard to extract satisfied ROIs from medical images. In order to overcome the shortcomings in segmentation from the V-Net model or the level set method, we propose a new image segmentation method, the VLSM-Net model in which these two methods are combined in this paper. Specifically, the V-Net model first uses the V-Net model to segment the ROIs, and sets the segmentation result as initial contour. It is then feds to the hybrid level set method for further fine segmentation. While I believe this topic is of great interest to our readers, I think it needs major revision before it is ready for publication. So, I recommend this manuscript for publication with major revisions.

1. In this manuscript, the authors did not explain the importance of CT Image in the introduction part. The authors should explain the importance of CT Image.

2) Title: The title of the manuscript is not impressive. It should be modified or rewritten it.

3) Correct the following statement “The experimental results conducted in the public datasets LiTS and LUNA show that, compared with the V-Net model or level set method alone, the sensitivity, precision and Dice coefficient values (DCV) of the VLSM-Net model are greatly improved by our model, revealing the validation of our model in 3D image segmentation”.

4) Keywords: The keywords should be small. So, modify the keywords.

5) Introduction part is not impressive. The references cited are very old. So, Improve it with some latest literature.

6) The authors should explain the following statement with recent references, “The first term of the formula represents the region term, where m is the predefined lower limit of the gray level of the object”.

7) Please justify the following statement “LUNA16 is obtained from the LIDC-IDRI (Lung Image Database Consortium and Image Database Resource Initiative) dataset, which is the largest public lung nodules dataset”.

8) The author should provide reason about this statement “Therefore, the VLSM-Net model can achieve better performances by combining the structural features of the V-Net model and the LSM in LiTS dataset”.

9) Comparison of the present results with other similar findings in the literature should be discussed in more detail. This is necessary in order to place this work together with other work in the field and to give more credibility to the present results.

10) Conclusion part is very long. Make it brief and improve by adding the results of your studies.

 

11) There are many grammatic mistakes. Improve the English grammar of the manuscript.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Paper can be accepted in present form