Code for NURBS Fitting of a Monte-Carlo Simulation?

[connorhenley]

Hello,

I was looking to develop an algorithm for measuring the properties of a turbid medium using a "single Monte-Carlo" approach, where a surface is fit to a white monte-carlo output with muS = 1 and then it's scaled by muA and muS to fit to measurements (similar to Martinelli et al. 2011 "Analysis of single Monte Carlo methods for prediction of reflectance from turbid media"). Unfortunately I need to run simulations for a few different values of g and n, and it looks like VTS only stores a reference surface for g = 0.8 and n = 1.4 .

I had two questions:

1. Is the code for fitting a NURBS surface to a monte-carlo output publicly available?
2. More generally--is the sMC approach antiquated? I'm not a diffuse optics specialist. sMC seems like the perfect solution to my problem but I haven't seen many recent publications on the topic. If it is antiquated, is there a better approach that is similar in terms of computational speed?

Thanks!

-Connor

[hayakawa16]

Hi Connor,

Thank you for your questions.

1. Yes. If you clone the VTS respository (git clone https://github.com/VirtualPhotonics/vts.git), you can see the source code that generates the NURBS calculations as described in the Martinelli et al. paper. If you go this route, I can steer you to the files to look at.
2. No, many groups are still using the scaled MC approach because it provides very fast results.

Best,
Carole

[connorhenley]

Hi Carole,

Thanks for the fast response! It would be great to know which files to look at. I've cloned the repo.

-Connor

[hayakawa16]

Hi Connor,

Look at the file in vts/src/Modeling/ForwardSolvers/NurbsForwardSolver.cs and same folder NurbsGenerator.cs, NurbsValues.cs for the code. The folder vts/src/Modeling/Resources/ReferenceNurbs are the reference runs for the RealDomain and SpatialFrequencyDomain.

Unfortunately I'm not too familiar with the details of the code. Generally, the "basic scaled" (ref: Kienle and Patterson. ”Determination of the optical properties of turbid media from a single Monte Carlo simulation”, Physics in Medicine and Biology 41(10):2221-2227, 1996) worked okay (code in MonteCarloForwardSolver.cs and MonteCarloLoader.cs with reference run in Resources/ReferenceData). Martinelli noticed that if the reference result had adaptive binning (larger bins for distant rhos and longer times), the statistics improved and provided better results. He then applied Non-uniform Rational B-Splines to smooth the data.

I think others have generated updated Resources files and let the code remain the same. I need to check on the success of this.

Best, Carole

[connorhenley]

I found the code. This was very helpful, thank you!

[hayakawa16]

You're welcome!