Excellent question, as temperature results stand and fall with accurate boundary conditions! Wang et al. discuss them for thermal MRI head simulations extensively in their 2008 paper (doi:10.1002/jmri.21556). Apart from that, I would probably introduce a constant temperature boundary condition at any truncated surfaces of the human body (e.g. 37° where you cut off your model in the EM simulation), and also include the "headrest" of the coil. While there will be almost no power dissipated in it, it sure acts as a thermal insulator and neglecting it can lead to significant underestimation of superficial temperature increase.

Dear Andre

One more question; the slice thermal distribution in the head tissue follows the local SAR or the peak-spatial SAR1g distribution, of the same slice?

Dear Salah,

the temperature distribution will ususally not follow the SAR distribution at all, e.g. you will see a different distribution and also the maximum at possibly different positions that with the SAR distribution. Consequently, there is only a very superficial similarity between the distributions in the same slice. (e.g. see Wang et al. 2007 doi:10.1002/jmri.20977). I hope this answers your question.

I really appreciate your reply

My question was based on a simulation I did for head tissue without considering the internal boundaries and the blood interface, but the external boundaries only (Skin/Air), and at that time the thermal distribution was some how follows the SAR1g distribution.

Sincerely

Yes, 1g SAR and temperature can look similar, as the SAR averaging crudely mimicks the spatial lowpass filtering due to thermal effects in the body. It is important to remember that for very short heating times, i.e. time spans much smaller than the involved time constants of thermal conduction etc., temperature rise is directly proportional to SAR. So your result also depends on how long you are heating. The steady state will surely look different than any short-term heating.

Thanks Andre

Your explanation is great.

Sincerely

I'm always happy to help. If you have any more questions, just let me know!

Best,

Andre

One thing not adressed here is the effect of increased thermal diffusion through increases in perfusion in tissues. In addition, even at lower temperature increases, thermal boundaries are significantly affected by the impact of sweating. This effect has been modelled e.g. for hyperthermia therapy purposes

This is true, and certainly needs to be modeled. The impact of these effects is also addressed in the first paper I referenced above (Consideration of physiological response in numerical models of temperature during MRI of the human head, doi:10.1002/jmri.21556). However, they found that the "effect of temperature-induced physiological changes is negligible (0.01-0.02 °C)" for the head Birdcage coil at the SAR limit (3 W/kg). Do you have some additional references where significant differences in temperature were found when considering these physiological effects?

Not (yet) for MRI specific simulations. As indicated, these effects have been studied in view of hyperthermia/thermal therapy. Have to think about the source papers though.May also have been books. Will come back to this.

However, we will be obtaining improved information though using our (unique) new MR guided hyperthermia device. This will enable thorough studies into this. Hence, hopefully I can come back with improved data soon.... 

I like this discussion though, since temperature will increasingly be the base for limits, as e.g. suggested in the attached paper

http://www.ncbi.nlm.nih.gov/pubmed/23553588

Yes, SAR as a safety measure is only a (pretty bad) crutch. I think one of the major obstacles is the lack of convincing in vivo validations of temperature simulations at MRI power deposition levels. There have been some attempts (e.g. Oh et al. 2010, doi:10.1002/mrm.22174), but this is by far not enough. Also, most of the simulations I have seen are not really convincing due to the lack of appropriate boundary conditions. Especially the inclusion of clothing or close contact to the patient bed or coil headrests etc. seems to have a significant effect in my own simulations, yet this is rarely considered. The MRI community could surely benefit from the insights of the hyperthermia community.

PS: Thank you for the paper, this looks quite interesting!

Dears

I had many works on calculating temperature increase due to the mobile phone antenna EM radiation. But it seems it is not easy for the MRI case. Based on some manuals and published papers I found the following while setting the boundary conditions for the thermo-simulation. I am trying to calculate the temperature with high resolution FDTD griddling and using a super-computer (with Tesla GPU accelerator). The MRI coil is a 3T low-pass filter. For simplicity and with the EM simulation I set the boundary truncates the torso.

For the thermo-simulation I defined  three boundaries: to Background, to internal air regions and to blood. 

1. To the Background (External boundaries): I am using "Mixed" type boundary with Outside T = 0 and Heat Transfer = 6. This is a typical boundary to model convection cooling through air contact and perspiration [1]. 

2.To the Tissues with air, i.e., Esophagus lumen Pharynx and Trachea lumen (Internal boundaries): I am using Neumann with Heat Flux = 0.

3. To Blod, i.e. Artery and Vein (Blood interface): I am using Dirichlet with Outside T = 0. This is a typical boundary since the blood is continuously cooling the surrounding tissue due to perfusion

Do you think such thermo-boundaries definition is enough to compute the temperature increase using a FDTD Using Pennis with Steady State Solver?

[1] E. Neufeld

High resolution hyperthermia treatment planning

SWISS FEDERAL INSTITUTE OF TECHNOLOGY ZURICH, 2008.

I appreciate you very interesting and valuable discussions and answers.

You may want to have a look at the attached paper for the values we usually use.

Are you sure that you are using a neuman boundary with heat flux of 0 to internal air? This is effectively doing nothing.

Fiala (1999, 2001) has some work on the effects of clothing.

 Would the assumption of zero heat flux to the internal air simply treat this air as perfectly insulating? It at least sounds reasonable to me, as I don't imagine a lot of heat exchange occurs between the tissues and internal air.

And @Salah: Are you using a commercial solver or some self-built simulation program for the thermal simulations?

Dear Andre

I am using the latest version of SEMCAD X

Dear,

how to define the boundary conditions for the temperature distribution in the human head when the source of radiation mobile phone (open space, isotermal, adiabatic or combination of them )? The program is CST.

Dear Vladimir

Unfortunately, I have no idea about CST, I am using SEMCAD X.

Dear Salah,

If you can, please explain a little more how are boundary conditions defined in SEMCAD X. I would appreciate that because that can be helpfull for me.

Dear Salah,

Did you make simulation for the temperature distribution and can me tell more about temperature increases (range)?