=
=
=
feasible
kASDP
N
k
N
l
i
jkl
lk
K lk k
i
j
d Sxb
d
1
1
,
2
~
(x)
Inverse Planning – using
HIPO
) ,
, (
1
w t
r f
fw f
M
m
m m
=
=
=
Minimise
Hybrid Inverse Planning and Optimization Engine HIPO
(1) Karabis A, Giannouli S, Baltas D:
Radiotherapy & Oncology
, 76, 29, 2005
(2) Karabis A, Belotti P, Baltas D:
WC 2009
The MINLP problem is solved by
HIPO
by a
hybrid approach
:
•
A heuristic
combining
SA
&
scoring
for the binary part
b
•
A quasi-Newton
for the
continuous part
t = x²
(
optimisation, L-BFGS
)
• Make use of a
non-linear
description
of the
transition function Θ
= .
+ ( . )