System and method for radiation therapy treatment planning using a memetic optimization algorithm
View Patent ↗A method for the optimization of radiation therapy treatment plans is disclosed. The disclosed method is equally-applicable to robotic radiosurgery as well as other types of radiosurgical delivery, intensity-modulated radiotherapy (IMRT), volumetric modulated arc therapy (VMAT), and three-dimensional conformal radiotherapy (3DCRT). A population-based heuristic approximation is used to perform a global search, and subsequently, a deterministic local trajectory search is employed to further refine the initial solution.
1. A method of radiotherapy treatment planning, comprising:
determining a set of beam geometry and dosimetric parameters from a totality of beams;
calculating dose deposition coefficients responsive to the beam geometry and dosimetric parameters;
generating a set of points corresponding to a tissue volume, wherein each of the points corresponds to a place of interaction of a beam with a tissue volume;
identifying a solution subset of beams from the totality of beams;
evaluating the solution based on one of an objective, fitness, or cost function; and
activating a radiation source in accordance with the solution,
wherein the objective function comprises a quadratic function of the form:
F
fit
(
x
⇀
,
w
⇀
,
d
⇀
,
v
⇀
)
=
Θ
(
v
t
,
min
-
v
p
,
min
)
w
t
,
min
∑
1
n
t
Θ
(
δ
thres
,
min
-
∑
1
b
DDC
ij
x
j
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
min
)
2
+
Θ
(
v
t
,
max
-
v
p
,
max
)
w
t
,
max
∑
1
n
t
Θ
(
∑
1
b
DDC
ij
x
j
-
δ
thres
,
max
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
max
)
2
+
∑
l
=
1
org
[
∑
m
=
1
c
l
m
[
Θ
(
v
l
m
-
v
p
,
l
m
)
]
w
l
m
∑
1
n
l
Θ
(
∑
1
b
DDC
kj
x
j
-
δ
thres
,
l
m
)
(
∑
1
b
DDC
kj
x
j
-
d
p
,
l
m
)
2
]
.
2. The method of radiotherapy treatment according to claim 1 , wherein identifying step is performed in accordance with at least one of population-based heuristics, population-based meta-heuristics, memes, a local search, local learning procedures, or global learning procedures.
3. The method of radiotherapy treatment according to claim 1 , wherein the identifying step is performed in accordance with a population-based meta-heuristic global optimization including at least one of evolutionary computation, ant colony, swarm intelligence, neural networks, differential evolution, artificial life, cultural algorithms, harmony search, artificial immune systems, learnable evolution models, and tabu search.
4. The method of radiotherapy treatment according to claim 1 , further including:
selecting at least one parent candidate;
combining parent candidates to generate at least one offspring solution;
generating at least one random mutation solution;
applying a local or global learning procedure to at least one individual solution; and
selecting at least one survival candidate from at least one of an offspring solution, a mutation solution, or a learning procedure solution.
5. The method of radiotherapy treatment according to claim 1 , further including performing a local search in the solution subset.
6. A radiotherapy system comprising:
a planning apparatus comprising:
at least one processor; and
a computer-readable storage medium in communication with the at least one processor wherein the computer-readable storage medium comprises one or more programming instructions for:
determining a set of beam geometry and dosimetric parameters from a totality of beams;
calculating dose deposition coefficients responsive to the beam geometry and dosimetric parameters;
generating a set of points corresponding to a tissue volume, wherein each of the points corresponds to a place of interaction of a beam with a tissue volume;
identifying a solution subset of beams from the totality of beams; and
evaluating the solution subset based on an objective function; and
a radiation treatment apparatus adapted to receive the evaluated solution subset and apply radiation to a planning treatment volume in accordance with the evaluated solution subset,
wherein the objective function comprises:
F
fit
(
x
⇀
,
w
⇀
,
d
⇀
,
v
⇀
)
=
Θ
(
v
t
,
min
-
v
p
,
min
)
w
t
,
min
∑
1
n
t
Θ
(
δ
thres
,
min
-
∑
1
b
DDC
ij
x
j
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
min
)
2
+
Θ
(
v
t
,
max
-
v
p
,
max
)
w
t
,
max
∑
1
n
t
Θ
(
∑
1
b
DDC
ij
x
j
-
δ
thres
,
max
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
max
)
2
+
∑
l
=
1
org
[
∑
m
=
1
c
l
m
[
Θ
(
v
l
m
-
v
p
,
l
m
)
]
w
l
m
∑
1
n
l
Θ
(
∑
1
b
DDC
kj
x
j
-
δ
thres
,
l
m
)
(
∑
1
b
DDC
kj
x
j
-
d
p
,
l
m
)
2
]
.
7. The radiotherapy system according to claim 6 , wherein the one or more programming instructions for identifying is performed in accordance with at least one of population-based heuristics, population-based meta-heuristics, memes, local search, local learning procedures, or global learning procedures.
8. The radiotherapy system according to claim 6 , wherein the one or more programming instructions for identifying is performed in accordance with a population-based meta-heuristic global optimization including at least one of evolutionary computation, ant colony, swarm intelligence, neural networks, differential evolution, artificial life, cultural algorithms, harmony search, artificial immune systems, learnable evolution models, or tabu search.
9. The radiotherapy system according to claim 6 , wherein the one or more programming instructions further comprises:
selecting at least one parent candidate;
combining parent candidates to generate at least one offspring solution;
generating at least one random mutation solution;
applying a local or global learning procedure to at least one individual solution; and
selecting at least one survival candidate from at least one of an offspring solution, a mutation solution, or a learning procedure solution.
10. The radiotherapy system according to claim 6 , wherein the one or more programming instructions further comprises performing a local search in the solution subset utilizing at least one of conjugate gradient, a gradient method and linear programming.
11. A radiotherapy planning system comprising:
at least one processor; and
machine-readable media comprising a set of executable instructions adapted for execution on the one or more processors, the set of executable instructions for performing radiotherapy planning comprises:
receiving a first dose-volume constraint for a planning treatment volume;
receiving a second dose-volume constraint for tissue outside of the planning treatment volume;
determining a set of beam properties for a plurality of beams using a population-based heuristic approximation in accordance with the first and second dose-volume constraints;
refining the set of beam properties for the plurality of beams using a deterministic local trajectory search; and
generating a set of instructions to instruct a radiation treatment apparatus to apply radiation to the planning treatment volume in accordance with the plurality of beams having the refined set of beam properties,
wherein the set of executable instructions further includes instructions for evaluating a beam property of the set of beam properties for a beam of the plurality of beams using:
F
fit
(
x
⇀
,
w
⇀
,
d
⇀
,
v
⇀
)
=
Θ
(
v
t
,
min
-
v
p
,
min
)
w
t
,
min
∑
1
n
t
Θ
(
δ
thres
,
min
-
∑
1
b
DDC
ij
x
j
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
min
)
2
+
Θ
(
v
t
,
max
-
v
p
,
max
)
w
t
,
max
∑
1
n
t
Θ
(
∑
1
b
DDC
ij
x
j
-
δ
thres
,
max
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
max
)
2
+
∑
l
=
1
org
[
∑
m
=
1
c
l
m
[
Θ
(
v
l
m
-
v
p
,
l
m
)
]
w
l
m
∑
1
n
l
Θ
(
∑
1
b
DDC
kj
x
j
-
δ
thres
,
l
m
)
(
∑
1
b
DDC
kj
x
j
-
d
p
,
l
m
)
2
]
.
12. The radiotherapy planning system according to claim 11 , wherein the radiation treatment apparatus includes a control system adapted to receive the set of instructions and control the radiation treatment apparatus in accordance with the set of instructions thereby applying the radiation to the planning treatment volume in accordance with the set of instructions.
13. The radiotherapy planning system according to claim 11 , wherein the population-based heuristic approximation is a population-based meta-heuristic global optimization.
14. The radiotherapy planning system according to claim 13 , wherein the population-based meta-heuristic global optimization includes at least one of evolutionary computation, ant colony, swarm intelligence, neural networks, differential evolution, artificial life, cultural algorithms, harmony search, artificial immune systems, learnable evolution models, or tabu search.
15. A radiotherapy treatment planning system, comprising:
a processor means for executing a set of instructions for radiation treatment planning;
a storage means for storing the set of instructions;
a parameterization means for determining geometry and dosimetric parameters for a radiation beam directed to a treatment tissue volume;
a geometric point representation means for selecting a point of the treatment tissue volume;
an evaluation means for evaluating the geometry and dosimetric parameters directed to the point of the treatment tissue volume based on an objective function; and
a heuristic means for heuristically determining a treatment plan in accordance with the evaluation of the geometry and dosimetric parameters,
wherein the objective function comprises:
F
fit
(
x
⇀
,
w
⇀
,
d
⇀
,
v
⇀
)
=
Θ
(
v
t
,
min
-
v
p
,
min
)
w
t
,
min
∑
1
n
t
Θ
(
δ
thres
,
min
-
∑
1
b
DDC
ij
x
j
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
min
)
2
+
Θ
(
v
t
,
max
-
v
p
,
max
)
w
t
,
max
∑
1
n
t
Θ
(
∑
1
b
DDC
ij
x
j
-
δ
thres
,
max
)
(
∑
1
b
DDC
ij
x
j
-
d
p
,
max
)
2
+
∑
l
=
1
org
[
∑
m
=
1
c
l
m
[
Θ
(
v
l
m
-
v
p
,
l
m
)
]
w
l
m
∑
1
n
l
Θ
(
∑
1
b
DDC
kj
x
j
-
δ
thres
,
l
m
)
(
∑
1
b
DDC
kj
x
j
-
d
p
,
l
m
)
2
]
.
16. The radiotherapy treatment planning system according to claim 15 , further comprising a truncation means for truncating a beam of the treatment plan.
17. The radiotherapy treatment planning system according to claim 15 , further comprising a local search means for optimizing a beam of the treatment plan.
18. The radiotherapy treatment planning system according to claim 15 , further comprising a radiation means for irradiating the treatment tissue volume in accordance with the treatment plan.