Verification of Convolutional Neural Network Cephalometric Landmark Identification

Round 1

Reviewer 1 Report

The authors try to compare prospective oral surgeons that are currently dental students with oral surgeon residents. The groups are not the same size which is not common practice but acceptable. I fail to understand if there was an ethical review committee approval. The authors should be congratulated for writing a good manuscript that paves the way to further test.
I have two concerns with the article. The number of patients is too small only 10 subject. I recognise the amount of work behind this manuscript, measuring 21 points with 30 operators. Nonetheless I fail to understand why the inclusion of 10 first year orthodontic residents. 

Table 2 first column is not 1,2,3...42 it should be 1,1,2,2,3,3,4,4....21,21

Table 2 column "p" should be in the Spearman Correlation has it belongs to it

Author Response

Thank you for your time spent and incisive comments regarding our research into the growing use and field of artificial intelligence in medicine.

I think there has been some misunderstanding because only orthodontic residents, practitioners and technicians were included in this study and no participants came from the field of Oral Surgery. The comparisons within and between these groups were included as part of the scientific method of this study based on recommendations received from the FDA, as several products relying on this technology are currently undergoing review for approval. The present study was submitted for ethics approval to the institutional review board and it was deemed that the present research was exempt from such approval.

The number of "patients" (cephalometric x-rays) needed for this study was determined statistically to be 5 according to the large number of observers used. It was decided to double this number in order to strengthen the findings.

I have amended Table 2 according to the Reviewer's recommendations regarding both the numerical labeling (most left column), and the category of the r-value of the Spearman Correlation.

Thank you

Reviewer 2 Report

The title of the manuscript is appropriate and indicative of the material, which is contained in the manuscript. The abstract is formulated upon all the criteria of the journal and describes clearly the purpose, materials, methods, results and the conclusions. 

The introduction is clearly formulated, complete and very detailed. There is a clear statement of the objectives of the study. Material and Methods used in the study scientifically valid and technically are correct. The procedures are clearly presented. There is no plagiarism in the article. The statistical analysis is complete, the author describes in a correct way the found result. The results and data gathered in the study are presented in a clear and logical method. The discussion is concisely stated. It develops arguments and theories from evidence. The conclusion is clear, concisely presented and based on the results of the study and on the statistical analysis. In my opinion the article should be accepted.

Author Response

Thank you for your time spent and thorough review.

Like you, I look forward to continuing research using the latest technologies, and hope that this manuscript will promote such further studies

Thank you

Reviewer 3 Report

Report attached

Comments for author File: Comments.pdf

Author Response

Thank you for your time spent and for your comments regarding the manuscript we submitted using artificial intelligence as a diagnostic tool in Orthodontics.

With regards to the indications that aspects of the manuscript "Can be improved" I would like to respond as such:

The research design was based on similar studies using analogous forms of artificial intelligence which are referenced in this manuscript within a comprehensive bibliography.

The description of the methods employed to carry out this study were based as noted above and refined according to statistician recommendations and FDA requirements. Furthermore, an internal research committee reviewed the protocol prior to initiating the study.

Regarding the presentation of the results, a revision was made to the numerical listing of cephalometric landmark identification according to the recommendations made by an additional reviewer. The textual presentation of the results fully describe the contents of the accompanying tables and figures.

I hope these responses satisfy the incisive comments provided by this reviewer (3).

Thank you  

Round 2

Reviewer 3 Report

The authors have addressed points raised.