Next Article in Journal
Pericardial Calcification: An Uncommon Case with Intraventricular Extension
Previous Article in Journal
Ultra-High Contrast MRI: The Whiteout Sign Shown with Divided Subtracted Inversion Recovery (dSIR) Sequences in Post-Insult Leukoencephalopathy Syndromes (PILS)
 
 
Article
Peer-Review Record

Fat Fraction Extracted from Whole-Body Magnetic Resonance (WB-MR) in Bone Metastatic Prostate Cancer: Intra- and Inter-Reader Agreement of Single-Slice and Volumetric Measurements

Tomography 2024, 10(7), 1014-1023; https://doi.org/10.3390/tomography10070075
by Giorgio Maria Agazzi 1, Nunzia Di Meo 2, Paolo Rondi 2,*, Chiara Saeli 3, Alberto Dalla Volta 4, Marika Vezzoli 5, Alfredo Berruti 4, Andrea Borghesi 2, Roberto Maroldi 2, Marco Ravanelli 2 and Davide Farina 2
Reviewer 1:
Reviewer 2: Anonymous
Reviewer 3: Anonymous
Tomography 2024, 10(7), 1014-1023; https://doi.org/10.3390/tomography10070075
Submission received: 2 May 2024 / Revised: 14 June 2024 / Accepted: 25 June 2024 / Published: 28 June 2024

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

 This study details the repeatability and reproducibility of fat fraction % in prostate cancer patients with bone metastatic disease and aims to demonstrates that values obtained with a single-slice and a volumetric segmentation are similar. The manuscript is well-written and describes the methods and results comprehensively and clearly.

 A few minor points:

            1       Line 101: Remove AGraphic from ‘Figure 1. AGraphic’

2      Line 107: ‘Voxel outliers more than 3 standard deviations away from the…’

3.       Line 116, 122 and throughout: Check references superscripted throughout, eg study8 and Barwick et al8.

4.       Line 153: Remove space ‘had small lesions that  segmented’

5.       Line 162: Make it clearer what the means and differences are in Figures 2, 3 and 4. Assume it is the %FF, but it’s not clear.

6.       Line 177: ‘small lesions, and single slices.’

7.       Line 199: Abbreviation ‘mCRPC’ used for the first time.

8.       Line 207: Suggest ‘The image analysis of this prospective study approaches the topic of lesion quantization by assessing the inter- and intra-observer repeatability of FF%.’

9.       Line 222: Consider rephrasing ‘and we should prefer larger ones when possible.’ Finding and assessing smaller lesions is better for the patient. Perhaps ‘and we should choose to make measurements on larger lesions whenever possible.’

10.    Line 235: Remove thus and add the. ‘…not justifying in our opinion the use of a single central slice instead’.

11.    Line 237: ‘Finally, it would be recommended…’

12.    Line 256: ‘Our preliminary data showed in keeping with the literature that…’ Are you referring to preliminary data on the %FF here? Again, it is unclear and reads like you are looking at the ADC. Also make this clear in the Supplementary Material.

13.    Line 266: This point would benefit from a reference(s), ‘robust region-growing algorithms are able (or will be in the near future)’.

 

 

 

Comments on the Quality of English Language

Generally good. A few suggestions above.

Author Response

Thank you very much for your positive feedback on our manuscript. We are delighted to hear that you found our study well-written and that the methods and results were described comprehensively and clearly.

We’ve addressed all the points you highlighted and modified the manuscript with your suggestions as you will see.

Here below we address more specifically the major points:

  1. Line 162: Make it clearer what the means and differences are in Figures 2, 3 and 4. Assume it is the %FF, but it’s not clear.

Thanks for your comment.

The legend was unclear. The means are expressed as fat fraction percentages (FF%) and represent the average of the two measurements. These measurements were taken by either one operator (for intra-reader agreement) or two different operators (for inter-rater agreement). The differences are also expressed as percentages, representing the percentage differences between the two measurements.

13.Line 266: This point would benefit from a reference(s), ‘robust region-growing algorithms are able (or will be in the near future)’.

Thank you for your comment, we have modified the sentence adding: “robust region-growing or machine learning algorithms” and we’ve include a reference (DOI: 10.1007/s10334-021-00958-5).

 

Best Regards, 

The authors.

Reviewer 2 Report

Comments and Suggestions for Authors

Whole-body MR imaging (WB-MR imaging) has become a modality of choice for detecting bone metastases in multiple cancers, and bone marrow involvement by multiple myeloma or lymphoma. This research  evaluate the concordance of FF% values obtained with a single-slice and a volumetric segmentation. General information about the 34 patients included in the study is recommended for analysis. The timeliness of reference is not good.

Comments on the Quality of English Language

There are no obvious language errors. 

Author Response

Dear Reviewer,

Thank you for your valuable feedback. We appreciate your considerations for our work.

Since this work focuses solely on the reproducibility of the fat fraction, we did not consider or have access to patient information, as we analyzed anonymized data. We apologize for not being able to address this point.

We know that the timeline of our reference is considering also papers that are older than 5 years, but since we are addressing an important topic that were not addressed before, we have tried to use only reference that we thought are more significative for our paper. We’ve added with this revision some new citations.

Thank you again for your insightful comments.

 

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript studied the repeatability and reproducibility of fat-fraction in WB-MRI of bone metastatic prostate cancer. The topic is of interest and has potential for clinical practice. However, the design of the study and the analysis of data should be improved. And there are a few points need to be addressed.

1. The patient number of 34 was chosen based on sample size calculation. However, this study is focused on repeatability and reproducibility analysis, more subjects is preferred.

2. Figure 1 is confusing. And it should be clarified how the lesion selection and volumetric segmentation are performed.

3. It seems that the difference of FF% values results from the disagreement of lesion segmentation. How the FF% is calculated in the lesion? And how automatic segmentation will affect this disagreement?

4. The correlation between the values of single-slice and volumetric measurements should be further explained. And why correlation is stronger for small lesions than large lesions?

Author Response

Dear reviewer, thank you for your valuable feedback.

Here there are point by point response to your kind suggestions, we have made modification of the paper according to your revisions:

  1. The patient number of 34 was chosen based on sample size calculation. However, this study is focused on repeatability and reproducibility analysis, more subjects is preferred.

Thank you for your insightful comment. We acknowledge that a larger sample size would enhance the robustness of our repeatability and reproducibility analysis. While our initial patient number of 34 was based on a sample size calculation that guarantee a statistical power of 80% (commonly used in literature), we understand the importance of including more subjects for more comprehensive results. We will consider this recommendation for future studies to improve the validity and reliability of our findings. We have a sentence addressing this matter in the limitation paragraph of our study.

  1. Figure 1 is confusing. And it should be clarified how the lesion selection and volumetric segmentation are performed.

“Volumetric segmentation was performed on axial sections (Figure 1)”... Volumetric segmentation was done drawing a polygonal ROI encompassing the entire lesion on every slice in which the lesion itself was visible.

Actually the measurements were done on 2D images, and thus the term “volumetric” might be considered improper (maybe it should be replaced by “multiple-slice 2D segmentation”). Nevertheless the term volumetric is generally accepted in these situations and using this software.

3. It seems that the difference of FF% values results from the disagreement of lesion segmentation. How the
FF% is calculated in the lesion? And how automatic segmentation will affect this disagreement?
The FF% is calculated in the lesion with the mathematical formula FF% = fat only/(water only+fat only)*100
Automatic segmentation would probably improve the agreement of calculated FF% because of reduction in
variability related to operator segmentation of lesions.
4. The correlation between the values of single-slice and volumetric measurements should be further
explained. And why correlation is stronger for small lesions than large lesions?
Correlation for smaller lesions is probably higher compared to larger ones because of possible
greater heterogeneity in larger lesions. This could influence results when comparing volumetric
measurement and single slice measurement. A further explanation could be related to added
segmentation variability when contouring a single slice comparing the segmentation of the whole
volume.

We hope to have addressed thorougly to your points, thanks for your precious suggestions,

the authors

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

There are still some issues about the revision.

1. Figure 1 is still confusing. In the caption, if I am correct, the left should be the FF without and with segmentation, and the right be 3D segmentation and that on the sagittal plane. Moreover, the 3D volumetric segmentation (upper-right) provide not much information and should be consider better presentation.

2. The FF% calculation in the lesion should be clarified. Is it based on the numbers of fat only voxels and water only voxels?

3. Discussion about how the lesion segmentation affects FF% values should be added.

Author Response

Dear Reviewer,

Thank you for your suggestions. We have addressed each of the issues raised in the reviewed manuscript as follows:

  1. Figure 1 Clarification:
    • We have revised Figure 1 adding letters to each panel and changed the caption to make it more clear.
    • The caption has been corrected to describe the contents.
    • We think that although the upper right image (c) does not provide much more information than the others, with this new description could be useful to have a picture of the 3D segmentation.
  2. Clarification on FF% Calculation in the Lesion:
    • We have added a detailed explanation in the manuscript to clarify how the FF% is performed. Specifically, it is now clearly stated that the FF% calculation is made using the T1-DIXON sequences and is based on the numbers of fat-only voxels and water-only voxels using the equation now added in the manuscript text.
  3. Discussion on Lesion Segmentation and FF% Values:
    • We have added to the discussion section  how lesion segmentation could be improved in the future using automatic segmentation. This addition could provide an important addition that might be included in future study to improve repeatability.

We believe these revisions address your concerns and improve the overall clarity and quality of the manuscript. Thank you again for your valuable feedback.

Best regards,

The authors

Back to TopTop