Numerical Assessment of the Risk of Abnormal Endothelialization for Diverter Devices: Clinical Data Driven Numerical Study

Round 1

Reviewer 1 Report

Authors proposed a numerical study of 4 cases of brain aneurysms assessing the endothelialization of diverter devices. Study is mostly presented as a prove of concept using common tools for segmentation and numerical simulations. Wall shear stress is reported as key parameter. Sample power remains the main limitation to bring strong evidence of the tested concept. Please consider following comments and suggestions:

1) Abstract: Please make sure to properly define abbreviations before using them.

2) Introduction and rational is mostly concentrated in WSS. A clear hypothesis is not set. A general aim is stated. Please consider rephrasing the aim for a more specific testing concept. Instead of "..., as well as to fine out the hemodynamic factors of abnormal stent endothelialization." specify what hemodynamic parameter you are studying. As currently presented, it is wrong to state that authors assessed several hemodynamic factors.

3) Provide a workflow figure illustrating all the process from the patient selection, scan, segmentation, mesh generation, simulation inputs and evaluated outputs.

4) Minor suggestion: avoid using references in figure legends. Cite within the manuscript. Otherwise, it may let to confusion regarding the origin of the reported data.

5) Further details in the selection of flow profile inputs is necessary. As current definition remains unclear.

6) Please improve figure quality, report proper scales, avoid unnecessary use of zeros (eg e+001), verify that all axis are easy to read when reading at 100% zoom, be consistent with units (eg. cms^-1 vs. cm/s). Most colorbars are unreadable. When zooming up, consider using using figure insets. Use properly panel letters and make sure you present all details in figure legends.

Author Response

Please see the attachment.

 

We thank the reviewer No-1 for the attention paid to our work and for making comments, which allowed us, in particular, to formulate the research hypothesis more clearly. In addition, we attach our responses to his comments.

Review report (Reviewer No-1)
Remark 1: Abstract: Please make sure to properly define bbreviations before using them.
Answer 1: Corrected 
Remark 2: Introduction and rational is mostly concentrated in WSS. A clear hypothesis is not set. A general aim is stated. Please consider rephrasing the aim for a more specific testing concept. Instead of "..., as well as to fine out the hemodynamic factors of abnormal stent endothelialization." specify what hemodynamic parameter you are studying. As currently presented, it is wrong to state that authors assessed several hemodynamic factors.
Answer 2: The formulation of the hypothesis was added in more clear format
Remark 3: Provide a workflow figure illustrating all the process from the patient selection, scan, segmentation, mesh generation, simulation inputs and evaluated outputs.
Answer 3:This figure is provided as a graphical-abstract
Remark 4: Minor suggestion: avoid using references in figure legends. Cite within the manuscript. Otherwise, it may let to confusion regarding the origin of the reported data.
Answer 4:Got it, references have been moved into the text
Remark 5: Further details in the selection of flow profile inputs is necessary. As current definition remains unclear.
Answer 5: The details are given in appendix 4. The link to the appendix is also given in the manuscript (see Methods section).
Remark 6: Please improve figure quality, report proper scales, avoid unnecessary use of zeros (eg e+001), verify that all axis are easy to read when reading at 100% zoom, be consistent with units (eg. cms^-1 vs. cm/s). Most colorbars are unreadable. When zooming up, consider using figure insets. Use properly panel letters and make sure you present all details in figure legends.
Answer 6: The number of decimal places has been reduced

Author Response File: Author Response.pdf

Reviewer 2 Report

This work is in line with the scope of the Journal of Personalized Medicine and shows the added value of numerical tools on medical planning. The work shows a good understanding of both, numerical tools and the medical problem.

However, the authors need to improve their work, especially the way it is presented as it becomes somewhat confusing. The main idea I got from reading your work is that you intend to verify the clinical outcome of an FDD application by using indicators from numerical simulation, for example, the WSS.

The authors are using a simplified approach, that was used before, but there are key points that are not presented. There is no information about the mesh, especially near the walls. Although the authors later refer to this as a limitation of the study, since your main indicator for the clinical outcome is the WSS, this should be clarified and given special attention. Additionally, the usage of porous media requires the specification of porous coefficients that will dictate the energy loss of the flow in that region. This value or consideration is not given. The authors should say how they calculated it and if it was in accordance with the type of mesh of the stent used in the application. Without these, the authors cannot specify the number of WSS and clinical outcomes.

In the Abstract, the authors refer to the numbers for MaxWSS of 1.23 and 1.68. Where do these values come from since this is the only time they appear? Are they yours? If so, they should be pointed out in the discussion and conclusion.

The line of thought in the writing of your paper should also be improved, as the why of including certain tests is not presented, or what was really done. For example, the authors point to the usage of both Newtonian and non-Newtonian viscosity models, but they say they chose a non-Newtonian model and in the results, it is not explicit what was really implemented.

P.S.: I'm annexing the PDF you submitted with comments on questions that have arisen during my reading.

Comments for author File: Comments.7z

Author Response

Please see the attachment

 

We thank the reviewer No-2 for the attention paid to our work and for detailed attention to each paragraph of our text. His comments made it possible to make the text more readable and to present the results of the study in a more convenient way. In addition, a number of technical errors and typos have been corrected. We attach answers to the list of his comments.

Review report (Reviewer No-2)

Remark 1: Explain where you get these values. Are these your results? Please explain further on the discution and present them in the conclusion. They only appear here.
Answer 1: This is the result we obtained, we added text into the “Results” section
Remark 2: A verb is missing. Maybe "are" is what you were trying to use?
Answer 2: Added
Remark 3: Please improve the caption of the table. Something like: Patients sample and treatment characterization, followed by the corresponding results.
Answer 3: Done
Remark 4: Table is not referred in the text before appearing
Answer 4: The table was moved down
Remark 5: For the four
Answer 5: Corrected
Remark 6: the considered
Answer 6: Corrected
Remark 7: Is figure 1 b) correspondent to this? Because it says anisotropic there.
Answer 7: Perhaps numbering of the patients and scenarios was disordered and confusing a bit- it has been changed
Remark 8: REFER on text
Answer 8: Done
Remark 9: C?
Answer 9: Y, corrected
Remark 10: to
Answer 10:Y, corrected
Remark 11: automating
Answer 11: Corrected
Remark 12: use sub-referencing in the figure such as a) b) and c) for easier reading In Figures 3 and 4.
Answer 12: Captions have been changed

Remark 13: The authors you refer say that "it is recommended to investigate different blood rheology models in IAs simulations when specific parameters to characterize the flow are needed, such as peak-systole WSS and OSI." But you are using this reference to refer only to non-Newtonian.
Answer 13: For the vessel itself (since it is much larger than 1 mm in diameter), it is sufficient to use the Newtonian fluid model - this is kind of a gold standard. For smaller vessels or stagnant motion, this is no longer true, and many authors (to which we referred, used non-Newtonian models). In our work, we did not make comparisons of different non-Newtonian models inside an aneurysm for a number of reasons. Firstly, we were more interested in the process of endothelialization on the stent, and not in the decrease in blood
flow velocity in the aneurysm, secondly, our colleagues conducted a comparative evaluation of the use of various models of non-Newtonian blood models for calculating the hemodynamics of aneurysms【Khe, A.; Chupakhin, A.; Cherevko, A.; Eliava, S.; Pilipenko, Y. Viscous dissipation energy as a risk factor in multiple cerebral aneurysms. Russ. J. Numer. Anal. Math. Modelling 2015, 30, 277–287.】and they did not find significant differences when
analyzing the results. The reason for using the mixed model refers to remark No. 26 and is given by us both here and in the text (updated).
Remark 14: From ANSYS Manual I can only verify the existence of this model "Casson". Please verify the name.
Answer 14:Done, Corrected
Remark 15: What are the values considered here?
Answer 15: We explained which values were used for different scenarios in the text.
Remark 16: was this zero pressure or zero relative pressure?
Furthermore, if you use pressures at the outputs, you cannot use velocities at the inlets or you would have an over defined problem.
Answer 16: Phrase was incorrect, corrected. About overdetermined problems and future plans, in fact we are going to solve the inverse problem. It’s ok to put P-values at the outlets (from the solver’s point of view), hence, of course we obtained an overdetermined problem and it is possible to obtain the case when the solution will not converge. So the way to solve it is :1) To calculate values of pressure near with the outlets, to put this values for the next
step of simulations and then to fit them so, in order to obtain pressure profile which we measured in the clinic (the point of the measurement is located in the internal domain of configuration, not at the outlet). We performed a preliminary simulation and it turns out that even when you vary pressure values about 0.5 mmHg it affects dramatically on the redistribution of blood flow in the whole configuration.
Remark 17: These t1, t2 ... Must be specified before presenting the table
Answer 17: Done
Remark 18: Did the simulation converge during the time steps? If not, the results cannot be trusted.
Answer 18: Of course all simulations converged. We have added the words about convergence criterion and also added a paragraph into appendix, dedicated convergence for the different approaches.

Remark 19: If you have the graph of velocities in Seconds, why use time step to show the results and not seconds as well?
Answer 19: We used confusing terms, corrected.
Remark 20: Split the results between the stationary and nonstationary cases as it is not intuitive when you transition from one type of simulation to the other.
Answer 20: Done
Remark 21: ANSYS
Answer 21: Corrected
Remark 22: to the vessel
Answer 22: Corrected
Remark 23: Was this truly experimental or numerical/modeled?
Answer 23: Everything is about the numerical simulations. We corrected this in the text.
Remark 24: The axis numbers are too small. This graph should also be better explained/discussed. Why is there such differences in near the boundary layer? WSS is highly dependent on the viscosity and velocity gradient near the wall. Assuming this error at the boundaries, there is no guarantee of the calculation precision. Furthermore, you are using a non-Newtonian model which varies with the shear rate, which is influenced near the wall.
Answer 24: We have corrected the graphic, the explanation of the reasons of these differences was already given, but we have added our thoughts why we think the simulation is correct.
Remark 25: Refer to Fig. 7
Answer 25: Was referred
Remark 26: Didn't the authors use non-Newtonian model? Or was it a mix of models? If the latter is the case, where is the reference to the simulation results or differences obtained?
Answer 26: In this paper we used Newtonian model for all configurations without stent (so-called initial configuration), and mixed model for the cases with stents, where Newtonian model was used for the vessel region and Non-newtonian for aneurysm sac. We add some phrases according to this remark in the text (Methods section)
Remark 27: Can't see the Legend without zooming and the scale is not the same. For comparison you should use same scale or justify the change in the scale.
Answer 27: We can’t provide the same legend for all the cases because the high variety of WSS values between patients and important points of WSS maximum will disappear according to this common scale for 3 of four patients. The quality of each subfigure is fine to zoom it and to see exact max and min values.
Remark 28: Is this a C?
Answer 28: No, we rebuilt the figures to make them more clear to understand.

Remark 29: Not referred in the text and caption is too confusing.
&

Remark 30: Is this true? If this caption is true, then there are missing figures. Or are you trying to say that this is the WSS variation?
Answer 29-30: We have rewritten the caption to the figure and have given a more detailed explanation for the quantity we are trying to describe.
Remark 31: This table is easier to understand if transposed.
Answer 31: We agree, done
Remark 32: What is the meaning of n2*8? Was it 8 time longer? If so, what is the main reason if the nodes are the same? I ask this because you refer on line 181 that you used a combined set of Newtonian and non-Newtonian models? This is never explicit referred and
generates doubts. Specially when you said before you used the Casson model.
Answer 32: Maybe it was not clear, that initial configuration means configuration without stent. So when we add any kind of stent, we change the amount of nodes as well as the number of iterations. Of course the number of nodes is slightly bigger than for configuration
without a stent, but the presence of FDD makes the problem dramatically complicated, that’s why there is such a difference. We have added a couple of words in the text, also absolute values for the scenarios with FDD replaced by relative values in %% to make the
table more clear for understanding.
Remark 33: Comparing scenario B and C from Tab. 3 it is possible to observe that the...
Answer 33: Agree, corrected
Remark 34: Verify this sentence and what the authors intend to express here.
Answer 34: Agree, very complicated, corrected
Remark 35: which configurations are you referring to? Stationary vs transient? Scenario B vsScenario C?
Answer 35: We undermine all stent scenarios (both: plug and tube), we agree it’s not obvious from the context, corrected.
Remark 36: specify this.
Answer 36: Corrected, look at previous remark
Remark 37: Is there any reason to change the order of patients? If the motive was the sorting of WSS variation values, you could say in the caption that it's ordered from higher to lower.
Answer 37: Exactly, we have specified the caption for the table
Remark 38: From where is this difference calculated? Additionlly, wasn't patient 1 a sucess? Why are you comparing with patient 2?
Answer 38: Explained and corrected
Remark 39: Show legend for the axis.
Answer 39: Done

Remark 40: Improve the title of the histogram. You are showing the prevalence of wss values after the neck by mesh cells.
Answer 40: Done
Remark 41: Transpose the table
Answer 41: Done
Remark 42: t4
Answer 42: Corrected
Remark 43: Didn't the authors used only a non-Newtonian model?
Answer 43: We used both steady and unsteady simulations, it was discussed in the previous remark
Remark 44: Appendix C
Answer 44: We tried, but some problem with the code. The code is exactly the same, I think technical support from the editorial board is able to help us
Remark 45: Is this letter C?
Answer 45: Images improved
Remark 46: Add that this is the stationary case
Answer 46: Done
Remark 47: This should be Presented at the beginning of the Annex to better understand its contents.
Answer 47: Done
Remark 48: Use the same legend in all cases that should be compared, i.e.: for Patiente 1, Patient 2, Patient 3 and Patient 4
Answer 48: Done

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All comments were addressed. Short comment. Please review reference formatting, a ? is shown along the manuscript.

Author Response

Please see the attachment.

 

We thank the reviewer No-1 for the attention paid to our work and for making comments. We attach our responses to his comments.
Review report (Reviewer No-1)
Remark 1: All comments were addressed. Short comment. Please review reference formatting, a ? is shown along the manuscript.
Answer 1: The second reviewer also highlighted this issue, but we checked the pdf and it was ok for all the references. We suppose it’s journal compiler malfunction

Author Response File: Author Response.pdf

Reviewer 2 Report

The given answers showed a good understanding of the problem and the changes helped to improve the more confusing parts of the paper.

I ask the authors to check again your references as the pdf I received has compilation errors, with the reference numbers appearing as "[?]".

Additionally, the authors should consider formating the legends of the results from scientific notation to decimal as the power of the results is only 10^1. I think this would make it easier for the reader to understand the numbers. Also, check the quality of the images in Appendix C as the compression seems to be blurring the legends.

Other than this, I don't find further improvements to be made.

Author Response

Please see the attachment.

 

We thank the reviewer No-2 for the attention paid to our work. We attach answers to the list of his comments.
Review report (Reviewer No-2)
Remark 1: I ask the authors to check again your references as the pdf I received has compilation errors, with the reference numbers appearing as "[?]".
Answer 1: We checked the manuscript but didn’t find any reference problems. May be it was compiler malfunction.
Remark 2: Additionally, the authors should consider formating the legends of the results from scientific notation to decimal as the power of the results is only 10^1. I think this would make it easier for the reader to understand the numbers. Also, check the quality of the images in Appendix C as the compression seems to be blurring the legends.
Answer 2: We remade figures in Appendix C, but about pictures captured from ANSYS - it’s default regime of picture saving, that is why we think it’s ok to read the pictures.

Author Response File: Author Response.pdf