Inquiry on Electron and Mass Density Mapping from CT in FLUKA

Dear FLUKA Support Team,

I am simulating dose distributions in FLUKA using CT images and would like to confirm whether my method for converting Hounsfield Units (HU) to material composition, mass density, and electron density aligns with FLUKA’s internal mechanisms.

1. HU-to-Material Mapping

  • HU ranges are mapped to predefined materials with specific elemental compositions and mass densities (e.g., for 700 ≤ HU < 800: H, C, N, O, Na, P, S, Mg, Ca. mass density: 1.460408 g/cm3).


  • Total Mass Density Calculation : For each voxel, linear interpolation is applied between adjacent HU ranges to compute the mass density (ρ):


    for example (for 700 ≤ HU < 800):
    image

  • Element Mass Fractions :
    Each material’s composition is defined by mass fractions (f.i) of elements(e.g., H: 0.061, C: 0.287). Element-specific mass densities are calculated as ρi (g/cm3) = f.i * ρ (g/cm3).

2. Electron Density Calculation

For each voxel, the electron density is computed using:

I would like to confirm whether this methodology is aligned with the way FLUKA internally interprets CT-based geometry when using ‘material.ini’ and ‘body.mat’.

Thank you for your time and support!

Dear @sy2119104

the files head.mat and body.mat you can find in the flair web site are not generic files to be used for any CT scan. Are only provided for demonstration purposes, one need to correct/adjust them for the CT scan they use.

The Conversion of HU to material and density is based on the Schneider parametrisation.

There is a discretisation applied on top so as each material to be used for HU in a range from 2/3*density to 3/2*density. This is due to the limitation of FLUKA on the number of materials to initialize. It is performed using the CORRFACTor card by +/-30% for the nuclear processes and electronic stopping power.

The calculation of the density you provided is correct. For the electron density is also correct if the additional factor WHAT(1) in CORRFACT for dE/dx process is 1.0 (which is the default in flair). Otherwise you need to multiply the scaled density with this factor to find out what fluka will use.