Two-View Mammogram Synthesis from Single-View Data Using Generative Adversarial Networks

Round 1

Reviewer 1 Report

The work constitutes an interesting exercise in the application of artificial intelligence methods for the reconstruction of mammographic images in the presence of a single projection. However, the actual usefulness and ultimate goal of the work seems unclear. To complete the work, a clinical investigation would be needed (for example an evaluation of the diagnostic potential of the reconstructed images by radiologists) and/or a quantitative validation (for example reconstruction and quantitative evaluation of images acquired with phantoms for quality control). As an alternative to such insights, I suggest underlining the preliminary and exclusively technical nature of the work in the introduction, and deepening it in the discussion and conclusions.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

A well-written article about using GAN to synthesize CC-view mammograms from MLO view data. I encourage the authors to proceed with further evaluations. I believe that the paper needs the following revisions:

·       Abstract: The aim of the paper should be clearer; I think that a few lines about the purpose of the provided work should be introduced in this section.

·       Results: I believe you should explain more and make it clearer what you meant by failed examples. If you could re-arrange the first paragraph in section 3.3, where you explain the definition and the difference between the failed and successful examples first, and then continue the analysis of the results.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf