Fluence Normalisation

121 views
Skip to first unread message

Ife E

unread,
Dec 6, 2023, 6:41:40 AM12/6/23
to mcx-users
Hi Dr. Fang,

Thanks for your help in my previous questions about MCX's treatment of fluence rate and units.

From what I understand when the normalisation flag is set the fluence rate output by the sim at each voxel should be scaled such that the source fluence rate is normalised to 1. So for a pencil beam it should be dirac(r-r0,t) with and amplitude of 1.

I've been using the sim data to validate my model for diffuse optical tomography and my results so far have been good in terms of localisation when solving for source position or fluence in the volume, but the scale of my reconstructed data is off by a few orders of magnitude (10^9). 

Looking at the data I'm getting from MCX cloud the fluence seems extremely large for normalised data, my largest value is 1.78e10! My initial instinct is that it is normalised but in units of 1/m^2/s, but I was hoping you could clarify more.

If it is an issue with my input json I ran the same simulation parameters that I've been using in MCX-cloud, Job ID: 18935558C927866A3D813AE8560764BA161F93F9.

Thanks,
Ife

Qianqian Fang

unread,
Dec 6, 2023, 5:56:24 PM12/6/23
to mcx-...@googlegroups.com, Ife E

hi Ife,

see my replies below.


On 12/6/23 06:41, Ife E wrote:
Hi Dr. Fang,

Thanks for your help in my previous questions about MCX's treatment of fluence rate and units.

From what I understand when the normalisation flag is set the fluence rate output by the sim at each voxel should be scaled such that the source fluence rate is normalised to 1. So for a pencil beam it should be dirac(r-r0,t) with and amplitude of 1.


this is incorrect. normalizing the solution does not mean the source fluence is 1.

also, direc() function has an amplitude of infinity, not 1 (it's integration is 1).


please read about Green's function to understand mcx's normalization

https://en.wikipedia.org/wiki/Green%27s_function


the normalized solution assumes the source term (usually the right-hand-side) of the PDE to be a delta function. this DOES NOT imply the solution (fluence-rate) should be 1 or infinity near the source location.



I've been using the sim data to validate my model for diffuse optical tomography and my results so far have been good in terms of localisation when solving for source position or fluence in the volume, but the scale of my reconstructed data is off by a few orders of magnitude (10^9).


check your output type - do you expect fluence-rate or fluence? by default, mcx output fluence rate (1/mm^2/s), but if you need fluence, you should multiply cfg.tstep, which is usually on the scale of 1e-9.

please see our sample code when comparing with analytical diffusion model

https://github.com/fangq/mcx/blob/master/mcxlab/examples/demo_validation_homogeneous.m#L116-L117



Looking at the data I'm getting from MCX cloud the fluence seems extremely large for normalised data, my largest value is 1.78e10! My initial instinct is that it is normalised but in units of 1/m^2/s, but I was hoping you could clarify more.

If it is an issue with my input json I ran the same simulation parameters that I've been using in MCX-cloud, Job ID: 18935558C927866A3D813AE8560764BA161F93F9.

Thanks,
Ife

--
You received this message because you are subscribed to the Google Groups "mcx-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mcx-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mcx-users/989e8b5b-533e-4d65-829b-01d31152df3cn%40googlegroups.com.
Reply all
Reply to author
Forward
0 new messages