Improving Localization of Deep Inclusions in Time-Resolved Diffuse Optical Tomography

Round 1

Reviewer 1 Report

In this manuscript Orive-Miguel et. al. have presented a review and proposed a new method to compute efficiently a long set of temporal windows in order to perform diffuse optical tomography. The research idea is highly valuable but, there some serious flaws that need to be addressed.

To start with:

a) An important paper Biomed Opt Express. 2013 Apr 1; 4(4): 569–583 was not cited. They had already proposed the methods mentioned in this paper and only difference being that they proposed inclusions at the depth of 1.5 cm but, in this paper Orive-Miguel et. al. have been proposed improvement beyond 2.5 cm.

b) In spite of the remarkable proposed improvement over what was proposed in Biomed Opt Express. 2013 Apr 1; 4(4): 569–583, this paper falls short as they have not provided any experimental evidence to support their theory. You can use data sets from previously conducted experiments (if you can find one) that would help prove improvement in resolution using your proposed equations.

c) Lastly, I could not find the algorithm or a flowchart that would help an user upgrade their existing software. The manuscript reads incoherently with bunch of equations. This problem would be automatically resolved once the authors try to implement their equations for real and show that it really works. This will also help them fine tune their manuscript such that this science is useful to people who are using this technique.

Overall, this is an excellent effort and I look forward to read this paper in future. This paper will attract lot of audience provided the authors' can show experimentally that their equations really work. There is a good chance that they might have to tweak their equations significantly while applying to real data set and I'll be happy to see those changes.

Author Response

Reply included in the PDF file.

David

Author Response File: Author Response.pdf

Reviewer 2 Report

 

Improving localization of deep inclusions in time-resolved diffuse optical tomography

David Orive-Miguel, Lionel Hervé, Laurent Condat, Jérôme Mars

 

The contribution provides an good overview on typical windows and their influence on reconstruction in terms of contrast, reconstruction accuracy and noise correlation in diffusive optical tomography. The summary is valuable for the reader to estimate effects of proper window selection on reconstruction. Standard moment window is used as comparison and for state of the art. Mellin-Laplace, Gaussian and Tuckey windows are considered with respect to their support in time and frequency domain, and their correlation matrices.

The overview on window types is interesting, but should be extended. Suggestion for improvement: Considering also windows derived from Generalized Gaussian distribution or B-spline based approaches.

A comparison in computational time / needed iteration numbers should be included as overview for all considered window approaches.

In Fig. 16f the phase term was not included, which was resulting in an underestimation of localization. Why was the phase term omitted? Vice versa, result for reconstruction based on only-phase would be perhaps interesting to consider. Phase information usually describes geometry and position and improves potentially the localization accuracy, but needs a careful handling in reconstruction.

Author Response

Dear Reviewer,

First of all, we would like to thank you for the feedback. We consider it very valuable for us and for the improvement of the paper.

In the next bullet points, you can find the answers to your suggestions:

We would like to emphasize that the aim of the paper is not only to provide an overview of state-of-the-art windows (e.g. standard and Mellin-Laplace windows) but also to propose a novel method from which a larger set of windows, such as Tukey and Gaussian windows, can be computed faster than previous methods (e.g. full-time direct model). We found the proposition of using Generalized Gaussian distribution very interesting. We did some reconstruction with different beta parameters (beta = 1, beta = 0.5) but reconstructions were very similar to the results obtained with Gaussian and Tukey windows. Nevertheless, we added a paragraph describing Generalized Gaussian distribution as a generalization of Gaussian and Laplacian distributions (see Subsection ‘Generalized Gaussian window’. In addition, ‘Gaussian window’ and ‘Exponenetial window’ sections were merged within that subsection). We agree that the computational time and number of iterations should be included in the paper. Therefore, we added a table with the computational time per iteration for each window and the full-time resolved case (see Section 3.1). We also added a figure with the number of iterations until converge is reached (see Table 2). Thank you for pointing out the omission of phase term. In practice, we found that to include the phase term makes the reconstruction more unstable. Although, some regularization could be applied we did not found a consistent and stable approach. For example, I tried some filters which include an additive term in the real part (to avoid a small denominator at phase computation) but results were not good enough. If the reviewer considers necessary to show results including phase term we are willing to try any other filters that the reviewer could suggest.

In addition to the above comments, some spelling and grammatical errors have been corrected.

We look forward to hearing from you in due time regarding our submission and to respond to any further questions and comments you may have.

Sincerely,

David Orive-Miguel

Reviewer 3 Report

See the attached file.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer,

First of all, we would like to thank you for your feedback. We consider it very valuable for us and for the improvement of the paper.

In the next bullet points, you can find the answers to your suggestions:

Regarding diffusion coefficient, thank you for pointing this out. We agree with this comment. Therefore, we have corrected the manuscript, added the suggested bibliography and redone the reconstructions (please, see line 31). Not significant differences are seen in the reconstructions since the absorption was several orders lower than the scattering term. Regarding the cardinality of a set, for countable sets it denotes the number of elements that it contains. In this case, it is the number of nodes within the region of interest. We have modified it to make it more clear to the readers.

In addition to the above comments, the spelling and grammatical errors have been corrected.

We look forward to hearing from you in due time regarding our submission and to respond to any further questions and comments you may have.

Sincerely,

David Orive-Miguel

Round 2

Reviewer 1 Report

The authors have made significant changes to the manuscript and have acknowledged the shortcomings in their study. I am satisfied with their answer and this manuscript needs to be accepted and published for others to try and test this algorithm. Congratulations!!!

Author Response

We would like to thank to the reviewer for the useful comments that he/she did. Those comments helped us to improve the quality of the paper.