Differences in dose calculation

Separate section for discussion on gamma transport
Post Reply
Ahmed_Naceur
Posts: 7
Joined: Tue Jun 16, 2020 10:08 pm
Security question 1: No
Security question 2: 7

Differences in dose calculation

Post by Ahmed_Naceur » Thu Dec 03, 2020 5:44 am

Dear Community,

I am currently running a photon transport simulation on a slab geometry (water [1-4] cm/bone [4-6] cm/lung [6-13] cm/water [13-31] cm). Here, gamma source is a linear source.

I tried three dose calculation methods with Serpent. The first using dr -200 (Serpent's dose rate), the second using dr -26 (total heating cross section) and the third using simultaneously dr -248 (for liquid water), dr -208 (bone cortical) and dr -228 (lung).

How units were managed:
1. I think that detector's response is D1= "Gy cm3/h". So I divided D1/volume/3600 = "Gy/s" ;
2. I think that detector's response is D2= "/s" (What I understand from the manual, may be I am wrong). So I divided (D2/Material_Density/Volume).*energy= "[(/s)*(cm3/kg)/cm3)]*J" = "J/Kg/s" = "Gy/s" ;
3. Exactly the same thing as in 1 for each material.

Questions:
Q.1. Can you, please, tell me if you agree with the mentioned units ?
Q.2. Can you, please, tell me why Methods 1 and 3 give a very, very, very close results (but not exactly the same), while there is a difference with respect to Method 2 (based on total heating cross section). These differences are important in the first water material and in the bone material. (Please, see Fig. 18MeV and become worse (catastrophic) for ultra-relativistic energy -- tested for 50 MeV and others -- ) ?
Q.3. All these methods will converge if the energy is decreased (please see Fig. 1MeV, become closer and closer for lower energies -- tested for lower energies --).
Q.4. If you want to give me some advice on the most fundamental method for dose calculations in Serpent, which one should I use and why ?

Note: Details on employed dose calculation methodology is very important for me. So if you want to add some details about how the dose rate is calculated by Serpent in each case, it is really appreciated.

Thank you so much for your time and efforts.
Attachments
Dose_detectors_results_18MeV.png
18MeV
Dose_detectors_results_18MeV.png (71.92 KiB) Viewed 604 times
Dose_detectors_results_1MeV_no_subivision.png
1MeV
Dose_detectors_results_1MeV_no_subivision.png (45.38 KiB) Viewed 604 times
-- Ahmed.

Ana Jambrina
Posts: 683
Joined: Tue May 26, 2020 5:32 pm
Security question 1: No
Security question 2: 7

Re: Differences in dose calculation

Post by Ana Jambrina » Fri Dec 04, 2020 5:23 pm

Some ideas here:
- Method 1/3:
What are the differences between method 1 and 3 (order of magnitude)?
What are the compositions/densities of those materials? (meaning, do they correspond with the pre-defined values in Serpent? - photon attenuation depends on nuclide/material composition and density).
How the detector output is obtained? (single run with multiple dr bins, or multiple runs with single dr bin - statistics’ effect).
- Method 2:
Energy deposition (W) —> Gy/s [x 10^3 / volume(cm^3)/rho(g/cm^3)].
For photon heating deposition there are two options: mt = -12, which is an analog estimate for the energy deposition by photons; or mt = -26, which uses KERMA coefficients. mt = -26 and mt = -12 should provide similar results in most cases, however, mt = -12 might be more accurate.
- Ana

Ahmed_Naceur
Posts: 7
Joined: Tue Jun 16, 2020 10:08 pm
Security question 1: No
Security question 2: 7

Re: Differences in dose calculation

Post by Ahmed_Naceur » Sat Jan 30, 2021 9:48 pm

Dear Ana,

Thank you so much for your response. I tested what you advised me to do.

Please, find attached to this message,
1. For 18MeV and 1MeV, the dose rate profile based on dr -200 [METHOD 1], dr -26 [METHOD 2], dr NIST -12 (What you advised) and dr -248 (for liquid water), dr -208 (bone cortical) and dr -228 (lung) [METHOD 3] ;
2. For 18 MeV and 1 MeV, the relative error from dr -200 ;

Method 1/3
As you can see, differences are not zero between Methods 1 and 3. I asked my question because I wanted to know how Serpent proceeds in each one of these 4 Methods, to choose the best for comparaison with Geant. Yes, If Serpent is using NIST densities and compositions, so this is the case for my benchmark. I will check deeply and may be we have very small differences.
Ana, it is a single run with multiple dr bins; so, there is absolutely no statistics' effect explaining the differences.
Overall, differences become higher and higher while increasing photon energy (tested until 50MeV).

Method 2
Sorry, I have made a mistake when typing my first message. Yes it is exactly the same units you mentioned (was watching the wrong bin's post-processing and writing). So I had the good units from the beginning.
When I tested dr mt=-12 (energy deposition), I saw that, at high energy, dr -12 is closer to dr -200 than the total heating cross section, i.e. dr -26. At low energy (please, see 1 MeV), the inverse happens. dr -26 became better. I am interested in low energy as well..

Your advice regarding the most fundamental method for dose calculations ?

My understanding of the philosophy behind each one of these Methods
If dose rate is evaluated using the detector integral (Manual, Eq. 7.1, p95), then
dr -200: attenuation_coeff*flux*dV*dE = cm^2/kg/gamma * gamma/cm^2/s * cm^3 * J =J/kg/s cm^3 = Gy cm^3/s. Please, if this is true, in this case, which attenuation coefficient (i.e. cross section) Serpent is using ?
dr -26 : instead of the attenuation coefficient, the macroscopic total heating cross-section is used, so I guess those cross sections of reactions involving heat generation like fluorescence and bremsstrahlung + energy deposition cross section ? For this reason, the dose is higher with this wrong method.
dr -12: instead of the attenuation coefficient, the macroscopic energy deposition cross-sections are used, those involving electrons generation like photo-electric effect, Compton and pair productions ? Electrons are deposited locally in Serpent (so this is energy)... (based on my library data, I guess)
dr NIST -248/-208/-228: what would it use from NIST exactly (should be just the material composition ?).


Thanks Ana for your time.
Attachments
s_18Me_Dose_differences.png
[18MeV-isotropic] Dose relative differences
s_18Me_Dose_differences.png (82.8 KiB) Viewed 501 times
s_1MeV_Dose_differences.png
[18MeV-isotropic] Dose relative differences
s_1MeV_Dose_differences.png (95.91 KiB) Viewed 501 times
s_1MeV_Dose.png
[1MeV-isotropic] Dose profile
s_1MeV_Dose.png (67.29 KiB) Viewed 501 times
s_18Me_Dose.png
[18MeV-isotropic] Dose profile
s_18Me_Dose.png (78.85 KiB) Viewed 501 times
-- Ahmed.

Ana Jambrina
Posts: 683
Joined: Tue May 26, 2020 5:32 pm
Security question 1: No
Security question 2: 7

Re: Differences in dose calculation

Post by Ana Jambrina » Sun Jan 31, 2021 5:38 pm

mt = -200 vs mt = -201/-248 responses:
- methodology is analogous
- In 'photon_attenuation.h' module you can check built-in mass-energy absorption coefficients (attenuation coefficients) - with reference Hubbell, J. H. and Seltzer, S.M. (2004), "Tables of X-Ray Mass Attenuation Coefficients and Mass Energy-Absorption Coefficients." (version 1.4). http://www.nist.gov/pml/data/xraycoef/
- mt =-200, attenuation coefficients are based on the material definition (nuclide composition and density); mt = -201 / -248, are based on pre-assigned material compositions.
- see the list of pre-defined material compositions list pre-defined materials. By using the command line option ‘-comp MAT [ID]' (see '-comp' command-line option, you can verify the pre-defined composition of material 'MAT' (further details: standard composition).

I would suggest to verify that the material compositions/densities are the same between the pre-assigned materials and your simulation, and check how you define the detectors (e.g. misinterpretation of detector response material and spatial domain of detector material)

mt = -12 vs mt = -26 responses :
in general, I would suggest using the analog energy deposition estimate, mt=-12: it should provide correct results within the limits of the accuracy of the physics models. In the other hand, mt=-26 provides the collision flux estimate of the energy deposition (or track-length flux estimate if ‘dtl’ detector definition is set) using average energy deposition data from the photon ACE library (a mismatch between the data and the photon models/simulations parameters mt = -26 could result in underestimating/overestimating the energy deposition; e.g. for high cut-off energy of photons mt=-26 would provide underestimate results). However, in case of need of improving statistics, mt=-26 seems to be the only choice (e.g. low density materials using delta-tracking detector ('dtl' option) or defining a minimum mean distance ('cfe' card).

[Having said that, just mentioned that mt=-200 (-201/-248), mt =-12 and mt=-26 not necessarily would provide same results: the methodologies are different as well as some data used on them].
- Ana

Post Reply