A Novel Polarized Light Microscope for the Examination of Birefringent Crystals in Synovial Fluid

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a very interesting approach to improve on some inherent problems with compensated polarised light microscopy for synovial fluid analysis. I have previously read a more technical oriented paper on SCPLM in the journal of biophotonics and I am delighted to see this diagnostic accuracy study. Although I think that the study is adequately designed and executed, the manuscript can be improved on clarity. 

Major comments

Overall: As this is a diagnostic accuracy study, I would recommend the authors to use the STARD checklist for reporting these studies. It is not required by the GUCDD author guidelines, however.

Introduction: This is not the only approach to improve on synovial fluid analysis. Previous work, for example the lens-free polarised light microscope (Nature scientific reports, 2016) should be cited. 

Methods: Paragraph 2.2: "Finally, SCPLM digital images have"; why is this? An image does not allow for analysis at different focal depths (as it is an image), while an operator using a CPLM can refocus and therefore look at different focal planes?

Methods: Paragraphs 2.5 and 2.6 are unclear and it is therefore difficult to understand your process. It is good that you used two independent observers and it seems that you have took great caution to reduce bias. It is unclear to me, however, how you calculated your performance measures. 

Methods: It seems that you calculate your performance measures per crystal instead of per FOV. As a FOV has several crystals, and mixtures of crystal types are quite rare, I think that this might influence certainty scores for individual crystals; e.g. when a small CPP crystal is next to a big one, one might be more certain of a CPP classification for the small CPP crystal. I would be interested to know the performance on a FOV level instead of a per-crystal analysis.

Results: I would recommend showing the 2x2 contingency tables used for calculating the sensitivity, specificity, and odds-ratio's.

Results: Figure 3: This figure is unclear and some parts are poorly aligned. I would recommend making a flow chart.

Discussion: The preselection of samples by the PI is a major limitation to the generalisability of the results and to the validity of the calculated specificity. For a diagnostic accuracy I would recommend a consecutive analysis of patient samples, including all (positive and negative) synovial fluids presented at a clinic within a predetermined period. 

Minor comments

Introduction: Malalignment -> misalignment 

Methods: Paragraph 2.2: "The Polarising CMOS sensor has [...]"; here you describe the size of the sensor, not the FOV under the microscope.

Methods: Paragraph 2.2: "See Error! Reference source [...]"; here something goes wrong with a cross reference (I think?).

Methods: Paragraph 2.2: "This sensor features 1600x [...]";  here you again describe the size of the sensor, not the FOV under the microscope.

Results: first paragraph: "This section may be divided [...]" 

Comments on the Quality of English Language

I would recommend a thorough read-through of the manuscript to check for some grammatical errors.

Author Response

Please see attached file.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

Interesting approach to potentially allow for a better method of crystal identification particularly for institutions without Raman Spectroscopy, but I agree with the discussion points that more studies need to be done and there should be a comparison with Raman spectroscopy to show the how well this method really works. The authors do not make clear how long the training period was for the raters. They only mention Zoom training sessions. Was it a few hours, or many hours, as that is also an important point whenever a new methodology is introduced.

In my reading of the paper, on Page 3 midway through the text there was a note stating "Error! Reference source not found". I am not sure if that is an error from the authors or the editors, and that is the only correction that I recommend.

Author Response

Comments 1: Interesting approach to potentially allow for a better method of crystal identification particularly for institutions without Raman Spectroscopy, but I agree with the discussion points that more studies need to be done and there should be a comparison with Raman spectroscopy to show the how well this method really works.

Response 1: Appreciate kind comments.

Comments 2: The authors do not make clear how long the training period was for the raters. They only mention Zoom training sessions. Was it a few hours, or many hours, as that is also an important point whenever a new methodology is introduced.

Response 2: We added approximate training length.  To section 2.5 “(roughly 30-minutes).”

Comments 3: In my reading of the paper, on Page 3 midway through the text there was a note stating "Error! Reference source not found". I am not sure if that is an error from the authors or the editors, and that is the only correction that I recommend.

Response 3: This was a broken link that has been removed and replaced with plain text, “Figure 2”.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for adapting your manuscript. I have no further comments.

Comments on the Quality of English Language

No comments.