Multi Simulation Platform for Time Domain Diffuse Optical Tomography: An Application to a Compact Hand-Held Reflectance Probe
, Antonio Pifferi 1,2, Simon Arridge 3, Fabrizio Martelli 4, Paola Taroni 1,2 and Andrea Farina 2Round 1
Reviewer 1 Report
The notion of this paper is interesting – the modelling of the electro-optical physical system in conjunction with the reconstruction algorithms is an important concept and useful to designers and engineers of the physical measurement instrument.
One of the issues in modelling an instrument is the huge number of potential configurations that are available to the designer. The other problem is the trading off of one limitation for another and how that influences the entire sensitivity and detectivity of the instrument. This paper aims to helping solve these conundrums for the designer.
However, as a designer of instruments, I feel that this paper is lacking in some important details (or perhaps they are not entirely clear in the manuscript).
1. The figure of merit is more complex than suggested. What is the inevitable final end-user’s (i.e. the physician’s) figure of merit? They would want to know detail of how likely it is to find (or miss) an inhomogeneity of a certain size with a defined contrast. How does the papers’ definition of the FoM connect with this ultimate goal?
2. The details of the physics of the noise and the definitions of the parameters could be better spelled out in detail. Perhaps a figure describing the measurement geometry could be included?
3. The discussion and the Figures 1 & 2 are confusing. It’s difficult to extract what is going on and there is little detail in presenting a figure of 20 plots, where one or two showing any contrast, and yet the real detail is hidden in the text of the plot legends and axis as the fonts are too small. A single colormap could cover each (or both) plots. Why is the width of the 5th column narrower than the others? – it covers the same physical extent.
4. Figures 3 & 4 have the same presentation problems as 1 & 2. The volumetric error in Figure 4 seems to get better then worse again as count rate increased. Is this because the error goes from -100% to +100% - what’s the difference between a negative error? Should this just be an error and presented as an absolute value?
5. Why the limitation on number of sources/detectors? How does the multiplexing of the sources/detector pairs relate to the measurement? This paper addresses some of the physical issues of TD-DOT but there are others – how long does a measurement take and how can the system designer trade this off with other constraints?
Is the matlab code going to be released? It would be difficult to trust the conclusions of this paper unless the code could be run and gives no real insight unless applied to my system with differing mechanical / electro-optical construction.
Comments for author File: 
Comments.pdf
Author Response
See attached file
Author Response File: Author Response.pdf
Reviewer 2 Report
This manuscript shows simulation results of a proposed tomographic imaging system based on time-domain diffusive optical tomography (TD-DOT). The development of new non-ionizing tomographic imaging system that can be used in the diagnosis of breast and thyroid cancer can produce significant advances in medicine and technology. I have several suggestions for modifications that I think can significantly improve the quality of the manuscript.
The authors need to provide clear information to the readers about the phantom system that they simulated to validate the method. They could do so by showing the spatial distribution of the image parameter (absorption, scattering) in cross-sections at different depths. That would be compared against the cross sections of those parameters that were calculated using the TD-DOT simulated system.
The authors use the Born approximation in the analytical study due to the higher speed of the calculation of that method. However, the authors need to justify in the paper why that approximation is accurate. If the method can accurately reproduce the image parameters in the validation that I suggested in the previous paragraph, this would be a good way to validate the approximation.
In Sec. 3 (Results and Discussions), the authors describe the results of simulations A, B, C, and D. It is not clear in the text how those relate to the Figures 1, 2, 3, and 4. The authors need to provide more information about these results. For example, the discussion on the simulation A, in which a study was carried out to determine the optimum voxel size given the dependence of the localization error with the voxel size. The authors should have provided a 2D plot showing how these two parameters are interrelated to facilitate the reproduction of the results by other researchers. The same procedure could be applied to the simulations B, C, and D.
The manuscript has an excessive number of paragraphs. For example, in the lines from 136 to 150 (15 lines) of the sub-section 2.2 (Tomographic kernel) there are five paragraphs. In the lines 286 to 303 (18 lines) of the Conclusion section there are also 5 paragraphs. Since many of those paragraphs have interconnected sentences, there should have been a smaller number of paragraphs in those portions of the manuscript.
Author Response
See attached file
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
I think the text in line 274 should read 10-1 rather than 10-1
Author Response
We thank the reviewer for the reporting. The typo has been corrected
Reviewer 2 Report
I thank the authors for addressing the suggestions and comments in my review. I noticed a few other items that could be improved in the paper. I am describing them below:
The authors should have highlighted in the revised version all the content that were added to the previous version of the paper in the present revision. Please, consider adopting that approach if another revision is needed.
In Figure 2, I did not see the red and the blue crosses addressed in the figure captions. That figure contains eight pink asterisks. Please, modify the figure or the figure captions for consistency.
The information on the phantom used by the authors to calculate the localization error in Fig. 2 was a helpful addition. However, the authors did not show any results of the actual imaging of the phantom in any of the positions considered. Even though the authors showed multiple images with different count rates and values of tau, it would have been even more valuable for the readers if the authors had shown cross sections of tomographic image of the virtual phantom obtained with this DO method, which could have been prepared from one of the cases with low localization error. For example, health professionals need to see the image of a tumor on the screen before they feel comfortable to diagnose the presence of a tumor. Just showing a low localization error does not show to the readers the full capability of this DO method. I strongly encourage the authors to include a tomographic image of the phantom obtained with the DO method that they proposed.
In line 64, the paragraph starting with “Moreover, in the last years an important …” could not have been a new paragraph, since the word “Moreover” indicates that this sentence is clearly a continuation of the previous sentence. The authors could have addressed that problem by removing the word “Moreover” from the beginning of that paragraph. I encourage the authors to carefully review the structuring of the paragraphs throughout the paper.
In lines 164, 165, you need to flesh out the meaning of the following sentence, since it was not properly formulated: “The two types of noises simulate at our best the experimental situation.”
Author Response
See attached file
Author Response File: Author Response.pdf
Round 3
Reviewer 2 Report
I thank the authors for addressing the concerns that I indicated.
