Mask Branch Network: Weakly Supervised Branch Network with a Template Mask for Classifying Masses in 3D Automated Breast Ultrasound

Round 1

Reviewer 1 Report

Please see below some comments, which I hope you will find useful to improve this work.

The literature article by Daekyung Kim et al. describes mask branch network: weakly supervised branch network with a template mask for classifying masses in 3D automated breast ultrasound. The simple summary is written in a still too technical language. I would suggest rewording some sentences, perhaps adding a very brief explanation. While I understand that rewording may be challenging. Please check the format of reference carefully. For example, line 316 and 328.

Is there any reason to use unknown [? ] in line 233,238 and 240. Please check carefully.

Author Response

We sincerely thank you for the advice about our manuscript. The comments provided us an opportunity to improve the expression of our paper. We respond to each comment as below.

Point1: The simple summary is written in a still too technical language.

• We appreciate for this pointer. As you mentioned, readers without a technical background my find it difficult to understand the summary. So we revised the abstract and added some sentences to improve the readability problem.

Point2: Please check the format of reference carefully.

• We have updated the supplementary material to fix the references: 'Screening for breast cancer in England' and 'Dedicated computer-aided detection software for automated 3D breast ultrasound'.

Point3: unknown [?] in line 233,238 and 240

• The unknown [?] was shown due to tex syntax error.We are sorry for this mistake, and will fix it.

Thanking again for this valuable feedback, we will reflect this in the final version.

Reviewer 2 Report

No comment

Author Response

We appreciate for reviewing with your positive opinion.

Reviewer 3 Report

The idea of the Mask Branch Network is very interesting and novel. However, it requires some additional input data coupled with the US images. The results are well presented and reported. However, in the introduction and abstract, the significance of the works is not clearly stated. This could be improved a bit. Here are some of my other comments.

1. The list of contributions at the end of the introduction is not very clear. Specially the third point needs rewriting.

2. Figure 2 has an image name as a type that needs to be corrected.

3, How the weights of the different losses are estimated needs a more elaborate explanation.

4. FIgure 3 is very hard to follow. Also, there are some [] symbols that are irrelevant I guess. The same case is in figure 5 as well.

5. Line 206, and line 238 are missing citations. Please correct them

6. Please also discuss the limitations of your works as well in the discussion part more clearly.

Author Response

We sincerely thank you for the advice about our manuscript. The comments provided us an opportunity to improve our paper. We respond to each comment as below.

Point1: In the introduction and abstract, the significance of the works is not clearly stated

• We appreciate for this pointer. As you mentioned, readers without a technical background my find it difficult to catch the importance of our method. So we have updated the abstract and added some sentences to improve the readability problem.

Point2: The list of contributions at the end of the introduction is not very clear. Specially the third point needs rewriting.

• We appreciate for this pointer. We have updated the third contribution point. For example, we have added "Extensive experiments exhibit that the proposed method contributes to improve the model’s performance" to the third one.

Point3: Figure 2 has an image name as a type that needs to be corrected.

• Thank you for pointing out missing details. We have updated the figure 2 to fix the name from 'mask branch' to 'mask branch network'.

Point4: How the weights of the different losses are estimated needs a more elaborate explanation.

• Thank you for your suggestion. We have added the explanation of uncertainty loss weight method as supplementary explanation.

Point5: Figure 3 is very hard to follow. Also, there are some [] symbols that are irrelevant I guess. The same case is in figure 5 as well.

• Thank you for your advice on Figure3. We have updated the caption of Figure3 to help readers to understand easier. We also have fixed some [] symbols.

Point6: Line 206, and line 238 are missing citations. Please correct them

• We are sorry for this mistake, and will correct the missing citations.

Point7: Please also discuss the limitations of your works as well in the discussion part more clearly.

• We appreciate for the advice on lack of discussion on limitations. We have added some scripts about our limitation of additional input information in the discussion section

Thanking again for this valuable feedback, we will reflect this in the final version.

Reviewer 4 Report

Dear Authors, 

please add scale to Fig 1

Please add some more references

Please elaborate more on Fig 2

 

Thank you

Author Response

We sincerely thank you for the advice about our manuscript. The comments provided us an opportunity to improve our paper. We respond to each comment as below.

Point 1: please add scale to Fig 1

• Thank you for your suggestion. We enlarged the size of Figure 1 to make visibility better.

Point 2: Please add some more references

• We appreciate for the advice on lack of references. We have added more reference paper: [2], [3], [4], [5], [8] and [9]

Point 3: Please elaborate more on Fig 2

• Thank you for your advice on Figure3. We have updated the caption of Figure3 to help readers to understand easier. We also have fixed some [] symbols.