I was recently asked to perform some simple tissue scattering simulations in the NIR, so I spent a little bit of time looking into the details, including reading the referenced Knowledge Base article.  I’m not sure why the calculated MFP and Transmission values do not agree with those used in the sample model.  Perhaps Csilla can respond.

However, in more general terms, I have a few thoughts for what they’re worth:

1. When implementing a scattering model in OpticStudio, it would seem to me that just the scattering coefficient (and not the absorption coefficient) is needed.  In the earlier literature, some authors use a so-called “total interaction coefficient” which is the sum of the linear absorption coefficient and the scattering coefficient.  For example, see this paper by Wang et al.:

This type of approach appears to be the origin for using the following equations in the KB article:

2. However, this “total interaction” approach seems to be geared toward pure Monte Carlo type models that do not utilize ray tracing.  Since OpticStudio traces scattered rays and determines the actual path length for each ray, it would be better to incorporate the absorption coefficient into the background transmission for the skin material. This can be done, for example, using a Table Glass that includes the refractive index and absorption.  Here is an example for 950 nm (with other wavelengths included just because OpticStudio prefers to have at least five wavelengths defined):

3. So, my approach is to just use the scattering coefficient alone to determine the scattering mean path and fold the transmission (absorption) effects into the Table Glass:

4. Lastly, it appears that the error bars on the published values for skin parameters are quite large.  So it’s not clear how well the model will mimic reality simply because the “real” skin parameters for any given situation may not be well known.  I found the following paper to be helpful:

It discusses various material parameter issues, provides tables of values for the parameters, and describes ray-trace scattering simulations.

Regards,

Jeff

I was recently asked to perform some simple tissue scattering simulations in the NIR, so I spent a little bit of time looking into the details, including reading the referenced Knowledge Base article.  I’m not sure why the calculated MFP and Transmission values do not agree with those used in the sample model.  Perhaps Csilla can respond.

However, in more general terms, I have a few thoughts for what they’re worth:

1. When implementing a scattering model in OpticStudio, it would seem to me that just the scattering coefficient (and not the absorption coefficient) is needed.  In the earlier literature, some authors use a so-called “total interaction coefficient” which is the sum of the linear absorption coefficient and the scattering coefficient.  For example, see this paper by Wang et al.:

 

This type of approach appears to be the origin for using the following equations in the KB article:

 

2. However, this “total interaction” approach seems to be geared toward pure Monte Carlo type models that do not utilize ray tracing.  Since OpticStudio traces scattered rays and determines the actual path length for each ray, it would be better to incorporate the absorption coefficient into the background transmission for the skin material. This can be done, for example, using a Table Glass that includes the refractive index and absorption.  Here is an example for 950 nm (with other wavelengths included just because OpticStudio prefers to have at least five wavelengths defined):

 

3. So, my approach is to just use the scattering coefficient alone to determine the scattering mean path and fold the transmission (absorption) effects into the Table Glass:

 

 

4. Lastly, it appears that the error bars on the published values for skin parameters are quite large.  So it’s not clear how well the model will mimic reality simply because the “real” skin parameters for any given situation may not be well known.  I found the following paper to be helpful:

 

It discusses various material parameter issues, provides tables of values for the parameters, and describes ray-trace scattering simulations.

Regards,

Jeff

Thanks for your post.

Did  you find any improvement in the use of Table glass instead of transmission coeficient as you mentioned?

I am working with skin modelling but i have some difficulties in order to define the layers that takes into account, both abs and scattering coeficient. Thank you in advance

S.

@Sonia.Perez  I haven’t performed a careful comparison of the two approaches, so unfortunately I can’t answer your question.  However, if during your work you uncover some important differences, it would be nice if you could let us know.

Thanks,

Jeff

Dear all, we have been spenting time to study how to model human skin for PPG in transmittion model instead of reflective one that has been used in  https://support.zemax.com/hc/en-us/articles/1500005579202-How-to-model-the-human-skin-and-optical-heart-rate-sensors-in-OpticStudio

Unfortunatly, we dont understand very well the model henvey-g for transmittion instead of reflective case in order to know how to define transmittance without interfering with “transmition parameter” or if henvey-G is right in that case instead of example link. 

If anyone can help for the future we will appreciate it.

thanks a lot

Hello All,

Sorry for the late reply, unfortunately, I didn’t see this post before.

@Chandan.Maurya Regarding your initial question about the differences in the numerical values, that is due to taking into account the relative blood content of the tissue layers. In the sample model, we took the blood content of the different skin layers into account by calculating the weighted average of the optical parameters of the blood and the surrounding tissue. If you double check, your numerical values for the living epidermis match exactly the ones in the model, because there is no blood content in that layer. I think you simply calculated your values for each layer based only on the tissue layer parameters, whereas we considered the parameters of the blood as well.

@Jeff.Wilde you are correct with your interpretation. The idea of using both the scattering and absorption coefficients to calculate the mean free path, and then using a Transmission factor smaller than 1, comes from Monte Carlo simulations. Indeed, OpticStudio can take into account absorption during ray tracing more accurately, if we define the absorption coefficients as material transmission parameters. As the model in the knowledgebase article was built for demonstration purposes, we started with the simpler Monte Carlo based approach, however your suggested method could be a more advanced representation of a skin model in OpticStudio.

@Sonia.Perez I think that the Henyey-Greenstein scattering model can be used for describing the human tissue in both reflective and transmissive PPG applications. However, please keep in mind that the applied wavelength are typically different in the reflective and transmissive cases.

Best,
Csilla

Thanks Casilla for your help,

in my study case, i have used a transmittance model that depens with the wavelength too. I wonder if absorption coefficients interfere with transmittance or if I must to consider both for the transmissive model. Must be the  absortion coefficient “1” when a transmittance model is consider in the material (GLASS TABLE)?

Thanks a lot

Hi @Sonia.Perez ,

As Jeff mentioned in his earlier post, there are two possible modelling approaches.

The one used in the knowledgebase article is inspired by Monte Carlo simulations, in which case a total interaction frequency is calculated based on both the scattering and absorption coefficients. This basically means that on average after propagating the mean free path, which is calculated as the reciprocal of the sum of the coefficients, there is going to be one interaction, which is either a scattering or an absorption event. So the Transmission parameter in this case defines the ratio of the scattering events compared to the total number of interactions, meaning if there is an absorption event, then the ray is fully absorbed, if there is a scattering event, the ray is scattered and preserves its energy. This is a simplified model, as the actual propagation path of the ray is not considered when calculating the absorption, rather a probabilistic approach is used.

On the other hand, if we take advantage of the ray tracing capabilities, it is possible to track the propagation path of each single ray, and calculate the absorption more accurately based on it. In this case, as the absorption is taken into account by the material properties, therefore the parameters of the Henyey-Greenstein model refer only to the scattering events. This means that the mean free path should be calculated based on solely the scattering coefficient, i.e. its reciprocal, and the Transmission parameter should be 1, meaning there is no energy loss during the scattering event itself.

If you used a Table Glass in your model, that means that you are using the second approach, which handles the two types of interactions separately. Absorption is taken into account by the material properties, and therefore the mean free path should be calculated as the reciprocal of the scattering coefficient, and the Transmission of the scattering model should be set to 1.

I hope this clears things up a bit, but if you have further questions, please ask.

Best,
Csilla