487 / 1224
3. Higher order of scatter
Instead of using h
1sc
with S
1sc
obtained from
D
prim
, to obtain D
1sc
couldn’t I distribute S
1sc
due to ALL orders of scattering
in a single step
using a different kernel, i.e. h
msc
…?
I could, but it is NOT a good idea due to:
•
h
msc
reducing less than h
1sc
with r
→
ray artefacts
•
accuracy close to boundaries of finite
geometries (discussed in the following)
Figures from: Carlsson & Ahnesjö Phys. Med. Biol. 45 357 (2000)