Thank you for your kind reply as always.
The one we used is Burker.
The method it describe was way too complex for me but the way it was done is very simple.
You segment out the 3 phantoms. then calculate the mean values. ( i mean automatically it is given just like in the segment statistics module)
Then we put the 3 values to the graph and calculate the equation for line with libreCalc,
So from the trend-line we can calculate the density of any know HU value.
As you can see the difference between 3D slicer and the burker is very small i guess…
Do you mean we take few more segments from each the 3 phantoms or we should use more phantoms with different densities ???