PowerPoint Presentation

481 / 1224

1. I need to be efficient
Implementation corrections:
• S1sc does not vary with θ within ΔΩ (Bθ,
bθ constant along transport line)
→ Kernel tilting problem less important as
E decreases
→ bθ evaluated recursively as a moving
average of previous and current step
• S1sc generated per unit r is constant
within the same voxel (scerma does not
vary considerably within voxels)
→ Scerma gradient problem more
important close to a source
→ in high scerma gradient regions scerma
is estimated piecewise from a log-linear
interpolation over r
θ1
θ2
θ3