Finding the Differences in Capillaries of Taste Buds between Smokers and Non-Smokers Using the Convolutional Neural Networks

Round 1

Reviewer 1 Report

Summary: The paper proposes a visual method to measure capillaries of taste buds with capillaroscopy and classified the difference between smokers and non-smokers using CNNa. The subject of the paper, evaluation, and findings are all very important to the intended population of the journal. The approach, evaluation method, and overall contribution are worth to publish.

• The use of three scaling factor at the same time on a CNN is both intriguing and promising
• The authors claimed that SSIM and SIFT scores are not sufficient for classifying difference between images in certain situations but it is not clear how the proposed model would compare against these models on a straight comparison

Originality/Novelty: Yes. The content of the paper, the methods, and the results are, to my knowledge, original and well defined.

Quality of Presentation: Overall, the article is well written. The figures and graphs are presented appropriately.

Interest to the Readers: The work is of interest to the readership of the journal.

Overall Merit: The work does contribute to the field and has benefits.

Author Response

1. Comment #1

The paper proposes a visual method to measure capillaries of taste buds with capillaroscopy and classified the difference between smokers and non-smokers using CNNa. The subject of the paper, evaluation, and findings are all very important to the intended population of the journal. The approach, evaluation method, and overall contribution are worth to publish.

-> Response #1

Thank you so much for you kind comment.  Visual data is enough easy to understand the abnormal detection regardless of the medical knowledge.

2. Comment #2

The use of three scaling factor at the same time on a CNN is both intriguing and promising. The authors claimed that SSIM and SIFT scores are not sufficient for classifying difference between images in certain situations but it is not clear how the proposed model would compare against these models on a straight comparison

-> Response #2

Thank you for your good comment. Table 3 represents the number of extracted featured parameters through CNN model. The minimum of number of parameters is over nearly 8,000,000, but the number of parameter through SIFT is about 8,000 as shown in Figure 7(b). The number of parameters through CNN model is approximately 1,000 times larger than that through SIFT. Thus, it is considered that it is necessary to have the enough large amount of extracted parameters for the image processing of capillaries.

Reviewer 2 Report

The paper presents an interesting application that could be useful in a real-life scenario. The following paper can be included in the paper

- Improved classification and localization approach to small bowel capsule endoscopy using convolutional neural network. Digestive Endoscopy. 2020 Jul 8.

Author Response

Thank you for your kind comment.

According to your suggestion, I will include the paper for the reference.

Reviewer 3 Report

The work is valuable but poorly presented! The reader (referee) is forced to strugle trough the text and its every sentence, starting from the title and abstract! The authors didn't even pay enough attention to compose a proper and precise title of their work!
The role of abstract is mixed with introduction. In introduction, the obviously good motivation for the topic is explained in clumsy sentences. Furthermore, in the introduction, there are parts that belong to conclusion, etc. Some things are repeated too many times, and some are not mentioned or are not clearly presented.

The authors should always think about how they present their thoughts to the readers.

The whole paper looks like it is written in a hurry, without proper check by (co)authors and without English proofreading. 
The scientific results should be presented more objectively and a little bit more self-critically. It seems that the authors lack the knowledge and skills of writing  a good scientific paper.

The recommendation is that this paper is accepted but after a serious overhaul, which can follow dozens of remarks written as  corrections in the attached pdf.

Comments for author File: Comments.pdf

Author Response

Thank you for your many comments.

It seemed that you did not understand our poor English Writing, but you gave us the correction with pdf file. Thank you so much. According to your suggestion, we revised our manuscript, thus it looks more improved, I think.

Sincerely