Multi-Granularity Dilated Transformer for Lung Nodule Classification via Local Focus Scheme

Round 1

Reviewer 1 Report

Thank you for the opportunity to read your research. My general comments are enlisted below:

1. Section 3.2 significantly relies on two already published works (36 and 37). Could the authors include the cited work as part of their illustration in Figure 1? This could help readers to understand your contribution. 

2. Section 4.2. Equations 9-13 are standard metrics in the field. I recommend omitting such detail. The mentions at the end of the first paragraph from Section 4.2 is sufficient.

3. In Section 3.2, the authors appoint two different works..e., 36 and 37; however, the authors decided not to include them in Table 1 (Comparison of state-of-art method). Is there a reason for this? If not, could the authors consider including works 36 and 37 for comparison purposes?

4. Could the authors consider updating the references? I found citations to work published from 2010. 

5. References 33 and 34 are identical. Revise and amend as required.

6. There are several papers with a citation from arXiv (e.g., 23, 36, 39), which do not always imply peer-review work. As this is not always the case, the authors should update their citations using the paper’s journal reference rather than the arcade’s reference.

 

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Reviewer 2 Report

 

 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Reviewer 3 Report

Detailed comments are given as follows:

1)      This paper proposes a novel Multi-Granularity Dilated Transformer with Local Focus Scheme to mine discriminative local details for reducing the ambiguity of highly similar nodules. The exprimental results also indicates that MGDFormer has a notable performance gain on hard samples compared with current methods.

2)      The paper is well-written, the discussions have verified the performance of each proposed modules, and the structure of paper is reasonable that I can follow it easily.

3)      There are still some aspects for improvement. The authors need to explain deformable dilated transformer in more detail, which can help us to understand the priciple of it. Moreover, the authors need to introduce the procedure of data pre-processing for more complete exprimental settings.

4)      What are the  pros and cons of non-separation method and your presented result in this paper?

5)      It is worth noting that the language in the paper needs to be carefully revised, with special attention to English grammar, typos, spelling and sentence structure, so that readers can clearly understand the objectives and results of the study.

 

6)      Conclusions should be updated with future research directions. 

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.doc

Round 2

Reviewer 2 Report

Authors have incorporated necessary corrections and also provided the required justification and clarifications. Article now appears better than the original submission