Detecting Breast Arterial Calcifications in Mammograms with Transfer Learning

Round 1

Reviewer 1 Report

In this manuscript, the breast X-ray dataset using BAC is divided into four severity levels, and its classification is improved through transfer learning to overcome the need for large training datasets. It also reduces the time required for model training and builds lighter, more accurate models. The research of the article is very interesting and practical. 

It is suggested that the author increase the programming environment used for training these models and the source of pre training models.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

1. Abstract is too lengthy. Please reduce the length and present in one paragraph.

2. Figure 1 - What are those yellow arrow referring to? Please clarify.

3. Reference style such as Mostafavi et al. (2015) [10] appears in Line 61 is incorrect. Please rectify. 

4. The paragraph appears in Lines 76 to 77 is too short. Please merge with other paragraph or elaborate it.

5. Please state the technical contributions of current work.

6. Please describe the main differences between the proposed work and those presented in Section 2.2 to highlight the uniqueness of current work.

7. Please present the confusion charts for 2-class and 4-class classification problems generated by all compared methods. 

8. Information in Table 2, particularly for "Optimizer" and "Loss Function" are incorrect. Please rectify this issue.

9. Table 4 - It will be good for authors to indicate the types of classification problems (e.g., 2-class, 4-class or etc) considered by each method presented in Table 4. 

10. Please ensure related ethic approval has been obtained to conduct the current research as it involves the medical imaging of hospital patients.  

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Authors have addressed all comments given properly. In my opinion, this paper is ready to be accepted for publication.