You are on page 1of 16

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/357898855

A Large-Scale Combinatorial Many-Objective Evolutionary Algorithm for


Intensity-Modulated Radiotherapy Planning

Article  in  IEEE transactions on neural networks / a publication of the IEEE Neural Networks Council · January 2022
DOI: 10.1109/TEVC.2022.3144675

CITATION READS

1 308

8 authors, including:

Ye Tian Chao Wang


Anhui University Anhui University
94 PUBLICATIONS   4,802 CITATIONS    16 PUBLICATIONS   276 CITATIONS   

SEE PROFILE SEE PROFILE

Xingyi Zhang Yaochu Jin


Anhui University Bielefeld University
186 PUBLICATIONS   6,855 CITATIONS    699 PUBLICATIONS   30,148 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Federated Deep Learning View project

Real-time control and optimisation in a Free Piston Engine Generator (FPEG) View project

All content following this page was uploaded by Ye Tian on 18 January 2022.

The user has requested enhancement of the downloaded file.


IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 1

A Large-Scale Combinatorial Many-Objective


Evolutionary Algorithm for Intensity-
Modulated Radiotherapy Planning
Ye Tian, Yuandong Feng, Chao Wang, Ruifen Cao, Xingyi Zhang, Senior Member, IEEE
Xi Pei, Kay Chen Tan, Fellow, IEEE, and Yaochu Jin, Fellow, IEEE

Abstract—Intensity-modulated radiotherapy (IMRT) is I. I NTRODUCTION


one of the most popular techniques for cancer treatment.
However, existing IMRT planning methods can only generate
one solution at a time, and consequently medical physicists
should perform the planning process many times to obtain
I NTENSITY-modulated radiotherapy (IMRT) is cur-
rently one of the most mature and widely used tech-
niques for cancer treatment. According to the clinical re-
diverse solutions to meet the requirement of a clinical
case. Meanwhile, multi-objective evolutionary algorithms
quirement of a cancer case, IMRT can provide acceptable
(MOEAs) have not been fully exploited in IMRT planning treatment plans based on multiple radiation beams with
since they are ineffective in optimizing the large number different shapes and intensities [1]. As plotted in Fig. 1, a
of discrete variables of IMRT. To bridge the gap, this paper linear accelerator can produce multiple beams by acceler-
formulates IMRT planning into a large-scale combinatorial ating electrons onto a target, where the emitted photons
many-objective optimization problem, and proposes a co-
evolutionary algorithm to solve it. In contrast to existing
are transmitted through a multileaf collimator (MLC)
MOEAs handling high-dimensional search spaces via vari- and form a therapeutic field [2]. The core issue in IMRT
able grouping or dimensionality reduction, the proposed planning is to obtain a therapeutic field by adjusting the
algorithm evolves one population with fine encoding for weights of beams and the positions of MLC leaves, so as
local exploitation and evolves another population with rough to maximize the tumor control probability and minimize
encoding for global exploration. Moreover, the convergence
speed is further accelerated by two customized local search
the normal tissue complication probability [3].
strategies. The experimental results verify that, the proposed The optimization of beam weights and MLC leaf
algorithm outperforms state-of-the-art MOEAs and IMRT positions is essentially a complex optimization problem,
planning methods on a variety of clinical cases. which contains many conflicting objectives and a large
Index Terms—Intensity-modulated radiotherapy, large- number of discrete variables (e.g., the test instances
scale optimization, many-objective optimization, combina- solved in Section IV contain five objectives and up to
torial optimization, evolutionary computation 3050 variables). This gained much research attention in
the last three decades, where a number of planning
Manuscript received –. This work was supported in part by the methods based on various optimizers have been tailored.
National Key R&D Program of China under Grant 2018AAA0100100, Earlier IMRT planning methods are mostly based on
in part by the National Natural Science Foundation of China under two-step optimization, where the fluence map optimiza-
Grant 61822301, Grant 61876123, Grant 61906001, Grant 62106002, and
Grant 62136008, in part by the Collaborative Innovation Program tion and the MLC sequencing are conducted in two
of Universities in Anhui Province under Grant GXXT-2020-013 and steps. The fluence map optimization aims to obtain the
Grant GXXT-2020-051, in part by the Research Grants Council of optimal beam weight of each grid in the discretized
the Hong Kong Special Administrative Region, China under Grant
PolyU11202418 and Grant PolyU11209219, and in part by an Alexander field, which is achieved by analytical algorithms (e.g.,
von Humboldt Professorship for Artificial Intelligence funded by the Fourier transform [4] and Taylor series expansion [5]),
Federal Ministry of Education and Research, Germany. (Corresponding mathematical programming methods (e.g., quadratic
author: Xingyi Zhang.)
Y. Tian, C. Wang, R. Cao, and X. Zhang are with the Information programming [6] and conjugate gradient method [7]),
Materials and Intelligent Sensing Laboratory of Anhui Province, An- metaheuristics (e.g., simulated annealing [8] and genetic
hui University, Hefei 230601, China (email: field910921@gmail.com; algorithm [9]), or other techniques [10]. While all the
wangchao8@ahu.edu.cn; rfcao@ahu.edu.cn; xyzhanghust@gmail.com).
Y. Feng is with the School of Computer Science and Tech- grids can only have the same beam weight at a time,
nology, Anhui University, Hefei 230601, China (email: yuan- the obtained fluence map should be divided into sev-
dongfeng@stu.ahu.edu.cn). eral apertures by MLC sequencing methods [11]. As
X. Pei is with the School of Nuclear Science and Technology, Univer-
sity of Science and Technology of China, Hefei 230026, China (email: illustrated in Fig. 2, the fluence map consists of three
xpei@ustc.edu.cn). apertures with different shapes, while the beam weight
K. C. Tan is with the Department of Computing, The Hong Kong in each aperture is unique.
Polytechnic University, Hong Kong SAR (email: kctan@polyu.edu.hk).
Y. Jin is with the Faculty of Technology, Bielefeld University, Bielefeld The optimization of fluence map is relatively easy,
33619, Germany (email: yaochu.jin@uni-bielefeld.de). however, the subsequent MLC sequencing may highly
0000–0000/00$00.00 ⃝
c 0000 IEEE
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 2

Multileaf
collimator
(MLC) Radiation
beam
Multileaf
collimator
(MLC)

Irregular therapeutic field


Electron linear accelerator Intensity-modulated radiotherapy (IMRT)
Dose-volume histogram (DVH)

Multiple f1: ‹‹—†‘•‡‘ˆͳ × × × × × × × ×


interactions × × × × × ×
f2: ƒš‹—†‘•‡‘ˆͳ
f3: ƒš‹—†‘•‡‘ˆͳ × ×
× × ×
f4: ƒš‹—†‘•‡‘ˆʹ × × × × × ×
f5: ”‡•…”‹„‡††‘•‡‘ˆͳ × × × × ×
f6: ‘•‡Ǧ˜‘Ž—‡…‘•‘ˆͳ × × × ×
ĂĂ × × × × × × × ×

Radiation treatment Solution assessment Treatment planning

Fig. 1. General procedure of intensity-modulated radiotherapy (IMRT).

Fluence map Aperture 1 and metaheuristics, just to name a few [17], [18]. So far,
× ××××××× × × × ×××× × direct aperture optimization has been a mature technique
× ×× ××× × × × ×××× ×
Grid × × × × × ×××× × and adopted in commercial systems [19].
× × × × × × ×× × Nevertheless, IMRT planning in clinical practice still
× ×× ××× × × × ×× ×
× ×× ×× × × × ×××× × requires much human expertise and takes hours or
Therapeutic
field
× ×× × × × × ×××× × even days for one clinical case [20]. Once the doctor
× ××××××× × × × ×××× × decides the treatment protocol, the medical physicist first
builds an optimization model and obtains its optimal
Aperture 3 Aperture 2 solution (i.e., treatment plan) via a planning method,
× ××××××× × × ×× ××× × then discusses with the doctor the quality of the plan
× ×× ××× × × ×× ××× ×
× × × × × ×× × according to the dose-volume histogram. If the plan
× × × × × × × × is not clinically acceptable, the medical physicist must
× ×× ××× × × × ×× × build a new model and re-run the planning method.
× ×× ×× × × × × ×
× ×× × × × ×× ××× × This procedure will probably be repeated many times
× ××××××× × × ×× ××× × as shown in Fig. 1, which is relatively time-consuming
since only one plan is obtained at a time.
Fig. 2. An illustrative example of MLC sequencing.
On the other hand, IMRT planning is naturally a
multi-objective optimization problem since the radiation
deteriorate the effectiveness since complex fluence maps damages both tumor and healthy cells [20]. To ensure the
cannot be accurately divided into few apertures [12]. By minimum doses of tumor cells and limit the maximum
contrast, some other IMRT planning methods directly doses of healthy cells, multiple conflicting objectives
optimize the shapes and beam weight of each aperture, should be optimized simultaneously. While most opti-
which is known as direct aperture optimization [13]. mizers adopted in existing planning methods are only
While the optimization problem contains a large number for single-objective optimization, the multiple objectives
of discrete variables (i.e., MLC leaf positions) and con- have to be aggregated by different weights. To determine
tinuous variables (i.e., beam weights), it is ineffective to the optimal weights for the objectives, an optimizer
directly optimize all of them by using the optimizers at should be employed many times to obtain the optimal
hand. To address this dilemma, some methods limit the solution for each setting of weights [21], [22]. In short,
shapes of apertures to certain ones [12], some methods the optimization process is also inefficient since only one
optimize the MLC leaf positions by random search [13] solution is obtained at a time.
or enumeration [14], some methods iteratively add new Therefore, the efficiency of IMRT planning can be
apertures and optimize only the beam weights [15], [16], considerably improved by generating multiple diverse
and some methods optimize the beam weights and MLC plans at a time, where ample alternatives are provided
leaf positions by mathematical programming methods to alleviate the uncertainty in both modeling and weight-
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 3

ing. Although multi-objective evolutionary algorithms of voxel v in terms of the i-th grid in the therapeutic
(MOEAs) have been applied to a variety of applications field of a beam. Besides, the intensity yi of the grid is
with large numbers of decision variables [23], [24], few determined by the beam weights and MLC leaf positions
of them have been adopted in IMRT planning. This of all the apertures related to a beam:
is because existing MOEAs mainly focus on continu- ∑
ous optimization, whereas little attention has been paid yi = wj pij , (2)
to large-scale combinatorial optimization problems like j

IMRT planning. The discrete variables in IMRT planning where wj denotes the beam weight of the j-th aperture,
pose stiff challenges for existing MOEAs to find well- pij = 0 indicates that the i-th grid is shielded by the
converged and diversified solutions [25]. Hence, this MLC in the j-th aperture, and pij = 1 otherwise.
paper aims to improve the efficiency of IMRT planning Based on the doses of voxels in PTV and OAR, a
by customizing an MOEA, which can obtain a set of solu- variety of objective functions have been formulated to
tions that present diverse trade-offs between the multiple describe the treatment protocol [7], [14], [26]–[29]. This
conflicting objectives in IMRT planning. Specifically, this paper adopts the following four types of objective func-
paper includes the following three contributions: tions to be minimized, which are easy to calculate and
1) The IMRT planning is formulated into a large-scale contain just a few parameters that should be predefined
combinatorial many-objective optimization prob- by medical physicists.
lem. The problem definition is applicable to dif- Minimum Dose: This objective function is to ensure the
ferent clinical cases, where medical physicists only minimum dose of each voxel, which is defined as:
need to vary the parameters in the problems. 1 ∑
2) A coevolutionary algorithm is proposed for solving M inDose(x, V, dmin ) = (max{0, dmin − dv })2 ,
|V |
the above problem, termed MOEA/IMRT. In or- v∈V
(3)
der to handle the high-dimensional discrete search
where x is a solution determining the values of wj and
space, the proposed algorithm evolves one popula-
pij in (2). Besides, V denotes a region of PTV or OAR and
tion with fine encoding and evolves another popu-
dmin denotes the predefined minimum dose. Obviously,
lation with rough encoding, where the balance be-
the value of M inDose is zero only when all doses are
tween exploitation and exploration can be achieved
greater than or equal to dmin .
by the cooperation of the two populations. Besides,
Maximum Dose: This objective function is to limit the
two local search strategies are suggested to repair
maximum dose of each voxel, which is defined as:
invalid solutions and accelerate the convergence.
3) The effectiveness of the proposed MOEA/IMRT 1 ∑
M axDose(x, V, dmax ) = (max{0, dv − dmax })2 ,
is tested on six clinical cases of different can- |V |
v∈V
cers. According to the experimental results, the (4)
proposed algorithm has better convergence and where dmax denotes the predefined maximum dose.
diversity performance than state-of-the-art MOEAs Similarly, the value of M axDose is zero only when all
and IMRT planning methods. doses are less than or equal to dmax .
The rest of this paper is organized as follows. Section II Prescribed Dose: This objective function is to make the
gives the proposed optimization model, and reviews ex- doses of all voxels as close to the prescribed dose as
isting multi-objective IMRT planning methods and large- possible, which is defined as:
scale MOEAs. Section III describes the proposed MOEA 1 ∑
for IMRT planning. Section IV analyzes the experimental P reDose(x, V, dpre ) = (dv − dpre )2 , (5)
|V |
results. Lastly, Section V draws the conclusion. v∈V

where dpre denotes the prescribed dose, and the value of


II. M ULTI -O BJECTIVE IMRT P LANNING P reDose is zero only when all doses are equal to dpre .
A. Problem Formulation Dose-Volume Constraint: This objective function is to
limit the maximum doses of all voxels, which is defined
The core task in IMRT planning is to find the optimal
as:
beam weights and MLC leaf positions for all the aper-
tures, so that the resultant doses of voxels in PTV (i.e., 1 ∑
DV C(x, V, ddvc , r) = (max{0, dv − ddvc })2
the planning tumor volume consisting of tumor cells) |V | (6)
v∈V
and OAR (i.e., the organ at risk consisting of healthy · (max{0, dr − dv }),
cells) meet the requirement of a clinical case. The dose
dv of a voxel can be calculated by [7], [26] where ddvc denotes the predefined maximum dose, dr
∑ denotes the r-th largest dose in V , and the value of DV C
dv = Div yi , (1) is zero only when the number of doses greater than ddvc
i is less than or equal to r.
where D is the dose deposition matrices obtained in ad- Considering all the regions in PTV and OAR, there
vance [7] and Div denotes the dose deposition coefficient may exist dozens of objective functions for one clinical
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 4

case. While existing work aggregates all the objectives single-objective optimization problems [34], [35], and
into one objective [7], [14], [26], [29] or two objectives some work searches for the optimal weights via particle
[27], [30], this paper builds the following large-scale swarm optimization [28].
combinatorial many-objective optimization problem by Although the above methods involve the concepts of
aggregating only similar objectives: “multi-objective optimization” and “Pareto optimality”,
they can only find a single solution at a time. On the
Minimize f (x) = (f1 (x), . . . , f5 (x))
∑ contrary, a few MOEAs have also been adopted in IMRT
f1 (x) = M inDose(x, V, dmin ) planning to find multiple solutions simultaneously. In
V ∈P T V [30], a multi-objective particle swarm optimization algo-

f2 (x) = M axDose(x, V, dmax ) rithm was employed for the IMRT planning with two
V ∈P T V objectives. In [27], a classical evolutionary algorithm
∑ NSGA-II [36] was directly applied to the IMRT planning
+ DV C(x, V, ddvc , r)
V ∈P T V , (7) with multiple objectives and constraints. In [25], the
∑ performance of NSGA-II was enhanced by a tailored
f3 (x) = M axDose(x, V, dmax ) initialization strategy, where the initial solutions are
V ∈OAR
∑ generated by a gradient based algorithm.
f4 (x) = DV C(x, V, ddvc , r) As a consequence, existing IMRT planning methods
V ∈OAR can hardly strike a balance between efficiency and effec-

f5 (x) = P reDose(x, V, dpre ) tiveness. On the one hand, the methods based on single-
V ∈P T V objective optimization should be performed many times
where x is a solution containing all the beam weights to generate a set of diverse plans, which is very inef-
and MLC leaf positions, P T V is the set of all regions ficient in clinical practice. On the other hand, although
in PTV, and OAR is the set of all regions in OAR. The a few methods adopt classical MOEAs to find multiple
first objective f1 is the core objective, which defines the plans in a single run, their effectiveness has not been
minimum dose of PTV and ensures the effectiveness of fully verified via comparative studies. More importantly,
radiotherapy. The second objective f2 contains the maxi- no state-of-the-art MOEA has been adopted and no
mum dose and dose-volume constraint of PTV, which search strategy has been tailored for IMRT planning.
are aggregated into a single objective since they are To take full advantage of evolutionary multi-objective
positively correlated with each other. The third objective optimization in IMRT planning, this paper proposes a
f3 defines the maximum dose of OAR and the fourth new MOEA with several customized strategies, includ-
objective f4 defines the dose-volume constraint of OAR, ing a coevolutionary framework and two local search
and the last objective f5 defines the prescribed dose of strategies. In Section IV, some state-of-the-art MOEAs
PTV, which makes the doses of voxels uniform so that will be tested in the experiments, which are verified to
the plan is more practical. be underperformed by the proposed algorithm.
Given a treatment protocol, the medical physicist
should first define the parameters dmin , dmax , dpre , ddvc , C. Existing Large-Scale MOEAs
and r for each region V , then optimize the above prob-
The proposed optimization problem is characterized
lem via an IMRT planning method or other optimizer.
by many objectives, large number of decision variables,
In the next two subsections, some state-of-the-art IMRT
and discrete search space. After decades of development,
planning methods and MOEAs are reviewed.
MOEAs are able to handle many objectives via enhanced
dominance relations [37], diversity based selection [38],
B. Existing Methods for Multi-Objective IMRT Planning objective decomposition [39], or performance indicators
In spite of the multiple conflicting objectives in IMRT [40]. However, due to the black-box search paradigms of
planning, existing planning methods are mostly based MOEAs, most of them encounter difficulties in handling
on single-objective optimization algorithms due to their a large number of decision variables. To tackle the curse
fast convergence speed. While most planning methods of dimensionality of decision space, some MOEAs based
aggregate the multiple objectives into a single one via the on decision variable grouping, decision space reduction,
weights set by hand, some other methods focus on the or novel search strategies have been proposed in recent
adaptive detection of the optimal weights. For example, years [24]. The decision variable grouping based MOEAs
some work only searches for the optimal solutions for adopt the divide-and-conquer strategy that divides the
each single objective [31], some work samples several decision variables via different grouping strategies, such
sets of weights and obtains the optimal solution for each as random grouping [41], [42], differential grouping
of the aggregated objectives [21], some work aggregates [43], and variable clustering [44]. The decision space
the objectives via principal component analysis [32], reduction based MOEAs aim to directly reduce the num-
some work navigates over the Pareto front by interpola- ber of decision variables via different dimensionality
tion between pre-found solutions [22], [33], some work reduction methods, such as problem transformation [45],
converts the original problem into several constrained problem reformulation [46], and unsupervised neural
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 5

networks [47]. The novel search strategy based MOEAs Fine encoding
Rough encoding
x2 = 3 x4 = 5
suggest new reproduction operators to accelerate the
x1 = 4 × × × × × × × × x2 = 4 × × × × × × × ×
convergence, such as competitive swarm optimizer [48],
x3 = 3 × × × × × × x4 = 3 × × × × × × × ×
self-evaluation evolution [49], and generative adversarial x5 = 1 × × x 6 = 1 x1 = 3 × × × × ×
networks [50]. These MOEAs have shown obvious supe- x7 = 2 × × × x8 = 1 × × × × ×
riority over classical ones in solving a variety of large- x9 = 3 × × × × × × x10 = 3 × × × × ×
scale multi-objective optimization problems [51]. x11 = 3 × × × × × x12 = 2 × × × × ×
Nevertheless, they cannot be directly applied to IMRT x13 = 3 × × × × x14 = 1 x3 = 3 × × × × ×
planning since all the three categories of search strate- x15 = 4 × × × × × × × × x16 = 4 × × × × × × × ×
gies are tailored for large-scale continuous optimization, Beam weight: x17 = 0.8 Beam weight: x5 = 0.8

which are unsuitable for handling the discrete variables


in IMRT planning. More seriously, the discrete variables Fig. 3. An illustrative example of two encoding schemes used in the
proposed algorithm. A solution may encode dozens of such apertures.
in IMRT planning cannot be arbitrary due to the re-
striction of aperture shapes, which even makes existing
MOEAs difficult to generate valid solutions. To address As a consequence, most MOEAs are likely to converge
these issues, the proposed algorithm suggests a novel slowly or even get trapped into local optimums, since the
coevolutionary framework for IMRT planning. Different proposed optimization problem has a high-dimensional
from the above MOEAs handling a large number of discrete search space and massive flat regions. To im-
decision variables via variable grouping, dimensionality prove the exploration ability, the proposed MOEA/IMRT
reduction, or new reproduction operators, the proposed suggests a coevolutionary framework, in which two
coevolutionary framework handles the large number of different encoding schemes are used simultaneously.
discrete variables from a new perspective. More specif- Fig. 3 illustrates the fine encoding and rough encoding
ically, two populations with different encoding schemes used in the proposed MOEA/IMRT. As can be observed,
are evolved cooperatively, where one population with the fine encoding directly encodes the positions of all
fine encoding is responsible for local exploitation, and MLC leaves (i.e., x1 , . . . , x16 ) and the weight of beam
the other population with rough encoding is responsible (i.e., x17 ) for each aperture, while the rough encoding
for global exploration. In the next section, the detailed encodes the aperture shape (i.e., left x1 , up x2 , width
procedure of the proposed algorithm is elaborated. x3 , and height x4 ) and the weight of beam (i.e., x5 ). The
fine encoding is straightforward and has been used in
existing IMRT planning methods, which can represent
III. T HE P ROPOSED A LGORITHM any possible aperture shapes; by contrast, the rough
A. Encoding Scheme of MOEA/IMRT encoding can only represent rectangular shapes. On the
other hand, the fine encoding contains much more de-
To solve the optimization problem defined in (7), any cision variables than the rough encoding, which means
MOEAs for continuous optimization (e.g., the genetic that the former is harder to be optimized than the latter.
algorithm based on simulated binary crossover [52] and That is, a population with fine encoding has good
polynomial mutation [53]) can be adopted in theory, exploitation ability and is responsible for approximating
where the decision variables of MLC leaf positions only the Pareto front, and a population with rough encoding
need to be rounded to the nearest integers before func- has good exploration ability and can escape from local
tion evaluation. However, the proposed optimization optimums. Therefore, the exploitation and exploration
problem provides the following difficulties that hinder can be balanced by the cooperation of two populations
MOEAs from reaching the global Pareto front: with different encoding schemes, where the convergence
• Flat regions in landscape: According to the definitions of the population with fine encoding can be significantly
in (3)–(6), the landscapes of minimum dose, maximum accelerated by the evolution of the population with
dose, and dose-volume constraint contain large flat re- rough encoding. Fig. 4 plots the first two objectives
gions due to the operator max, making the proposed of the populations on the Esophagus dataset, where
optimization problem difficult to be optimized. the populations are collected at different generations
• Discrete variables: Since the proposed optimization of the proposed MOEA/IMRT with fine encoding and
problem has a discrete search space, the rounding both the two encoding schemes, and the experimental
of real variables will introduce many small flat settings are referred to Section IV-A. It is obvious that
regions in the landscape, and hence the evolution the algorithm with fine encoding converges quite slowly
of a population will become more challenging. after 10000 function evaluations, while the algorithm
• Restriction of variables: Due to the physical limitation, with both fine encoding and rough encoding can de-
the aperture shapes can neither be too small nor crease the objective values steadily and obtain much
disconnected, which means that many of the ran- better solutions at last. Furthermore, Fig. 5 depicts the
domly generated solutions are invalid and imprac- convergence profiles of hypervolume (HV) [54] values
tical. Hence, the search space becomes discrete even obtained by MOEA/IMRT with a single and two en-
if real variables are optimized instead of integers. coding schemes on the same dataset, averaged over 30
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 6

MOEA/IMRT with fine encoding


Algorithm 1: Procedure of MOEA/IMRT
100 evaluations

10 0 10 000 evaluations
100 000 evaluations
Input: N (population size), K (number of exchanged
200 000 evaluations solutions)
Output: P1 (final population)
10-2
1 P1 ← Randomly generate N solutions of fine encoding;
2 P2 ← Randomly generate N solutions of rough encoding;
f2

3 [P1 , P2 ] ← LocalSearchA(P1 , P2 );
-4
4 while termination criterion is not fulfilled do
10
5 P1′ ← Select N parents from P1 via binary tournament
selection;
6 O1 ← Generate offsprings based on P1′ ;
10-6
10
-4
10
-3
10
-2
10
-1 7 P2′ ← Select N parents from P2 via binary tournament
f1 selection;
MOEA/IMRT with fine encoding and rough encoding 8 O2 ← Generate offsprings based on P2′ ;
100 evaluations 9 [O1 , O2 ] ← LocalSearchA(O1 , O2 );
100
10 000 evaluations 10 [O1′ , O2′ ] ← Exchange(O1 , O2 , K);
P1 ← P1 ∪ O1 ∪ O1′ ;
100 000 evaluations
200 000 evaluations 11
12 P2 ← P2 ∪ O2 ∪ O2′ ;
-2 13 P1 ← Truncate P1 via environmental selection;
10
14 P2 ← Truncate P2 via environmental selection;
f2

15 P1 ← LocalSearchB(P1 );
10
-4 16 return P1 ;

10-6
-4 -3 -2 -1
having N solutions with fine encoding and a popula-
10 10 10 10
f1 tion P2 having N solutions with rough encoding are
randomly initialized, and the first local search strat-
Fig. 4. Comparison between the first two objectives of the populations egy LocalSearchA() is conducted on the two popula-
on the Esophagus dataset, which are collected at different generations tions to repair invalid solutions. At each generation of
of the proposed algorithm with fine encoding and both the two
encoding schemes. MOEA/IMRT, two mating pools P1′ and P2′ are first
constructed by selecting N solutions from P1 and P2 ,
1 respectively. Then, the mating pools are used to gener-
ate two offspring population O1 and O2 via simulated
0.95
binary crossover and polynomial mutation, and the off-
0.9
springs are also repaired by the first local search strat-
egy. Afterwards, in the operation of offspring exchange
0.85 Fine encoding Exchange(), some offsprings with fine encoding in O1
Rough encoding
Rough + fine encoding are converted into solutions with rough encoding and
stored in O2′ , and some offsprings with rough encoding
0.8
1 2 3 4 5
Number of evaluations 105
in O2 are converted into solutions with fine encoding and
Fig. 5. Convergence profiles of mean HV obtained by MOEA/IMRT stored in O1′ . Lastly, the population P1 is combined with
with a single and two encoding schemes on the Esophagus dataset. O1 ∪ O1′ and truncated by environmental selection, and
the population P2 is combined with O2 ∪ O2′ and also
runs. According to the figure, the algorithm with rough truncated by environmental selection. Besides, the sec-
encoding converges much faster than the algorithm with ond local search strategy LocalSearchB() is conducted
fine encoding since the rough encoding has a better on P1 for further enhancement of the convergence.
exploration ability. However, the algorithm with rough
encoding converges prematurely and finally obtains the For the selection strategies in the proposed
worst HV value, since it cannot represent irregular aper- MOEA/IMRT, the mating selection and environmental
ture shapes to further reduce the objective values. By selection in SPEA2 [55] with shift-based density
contrast, the algorithm with fine encoding obtains better estimation [38] are adopted for simplicity, which are
result due to its good exploitation ability. Therefore, by flexible to be embedded in different MOEAs and
adopting both the rough encoding and fine encoding, have demonstrated high effectiveness in handling many
the proposed MOEA/IMRT has good exploration ability objectives. For the exchange of offsprings, MOEA/IMRT
and good exploitation ability at the same time, leading converts between the solutions with fine encoding and
to the fastest convergence speed and the best HV value. rough encoding by the steps given in Algorithm 2.
Firstly, some solutions are separately selected from the
two offspring populations to be exchanged, which is
B. General Procedure of MOEA/IMRT achieved by the same strategy as the environmental
The procedure of the proposed MOEA/IMRT is pre- selection of MOEA/IMRT. Note that only K solutions
sented in Algorithm 1. To begin with, a population P1 rather than all the solutions are selected here, since
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 7

Algorithm 2: Exchange(O1 , O2 , K) Rough encoding


Fine encoding
y2 = 2 y4 = 6
Input: O1 (offsprings with fine encoding), O2 (offsprings x1 = 4 × × × × × × × × x2 = 4 × × × × × × × ×
with rough coding), K (number of exchanged x3 = 3 × × × × × × x4 = 3 y1 = 4 × × × × ×
solutions) x5 = 1 × × x6 = 1 × × × × ×
Output: O1′ (converted offsprings with fine encoding), O2′ x7 = 2 × × × x8 = 1 Convert × × × × ×
(converted offsprings with rough coding) x9 = 3 × × × × × × x10 = 3 × × × × ×
1 O1K ← Select K solutions from O1 via environmental selection; x11 = 3 × × × × × x12 = 2 × × × × ×
2 O2K ← Select K solutions from O2 via environmental selection; x13 = 3 × × × × x14 = 1 y3 = 3 × × × × ×
//Convert fine encoding into rough encoding x15 = 4 × × × × × × × × x16 = 4 × × × × × × × ×
3 O2′ ← ∅; Beam weight: x17 = 0.8 Beam weight: y5 = 0.8
4 foreach solution p ∈ O1K do Rough encoding
5 q ← ∅; y2 = 3 y4 = 5
Fine encoding
6 foreach aperture x ∈ p do × × × × × × × × x1 = 4 × × × × × × × × x2 = 4
7 Convert x into y by (8); × × × × × × × × x3 = 4 × × × × × × × × x4 = 4
8 q ← q ∪ {y}; y1 = 3 × × × × × x5 = 2 × × × × × x6 = 3
9 O2′ ← O2′ ∪ {q}; × × × × × Convert x7 = 2 × × × × × x8 = 3
× × × × × x9 = 2 × × × × × x10 = 3
//Convert rough encoding into fine encoding
× × × x12 = 3
O1′ ← ∅;
× × × × × x11 = 2 × ×
10
y3 = 3 × × × × × x13 = 2 × × × × × x14 = 3
11 foreach solution q ∈ O2K do
× × × × × × × × x15 = 4 × × × × × × × × x16 = 4
12 p ← ∅;
Beam weight: y5 = 0.8 Beam weight: x17 = 0.8
13 foreach aperture y ∈ q do
14 Convert y into x by (9);
15 p ← p ∪ {x}; Fig. 6. An illustrative example of the conversion between fine encoding
and rough encoding.
16 O1′ ← O1′ ∪ {p};
17 return O1′ and O2′ ; (x1 , x2 , . . . , xd ) with fine encoding by
⌈c⌉
x2i−1 = , i < y2 or i ≥ y2 + y4
the exchanged solutions need to be re-evaluated 2
and consume additional computational resource. A x2i−1 = y1 − 1, y2 ≤ i < y2 + y4
⌊c⌋
parameter sensitivity analysis of K will be performed x2i = , i < y2 or i ≥ y2 + y4 . (9)
in Section IV-C. After selecting some solutions from 2
the offspring populations, the encoding scheme of x2i = c − y1 − y3 + 1, y2 ≤ i < y2 + y4
each selected solution is changed. As illustrated in xd = y5
Fig. 6, the aperture shape is smoothed when a solution As a consequence, the proposed MOEA/IMRT sep-
with fine encoding is converted into rough encoding, arately evolves two populations with different encod-
while the aperture shape keeps unchanged when a ing schemes by using the same strategies, and reaches
solution with rough encoding is converted into fine cooperation between the two populations by sharing
encoding. Formally, for a solution p with fine encoding, all the offsprings. The proposed coevolutionary frame-
its decision variables x = (x1 , x2 , . . . , xd ) corresponding work is different from existing ones in terms of two
to each aperture are converted into y = (y1 , y2 , . . . , y5 ) aspects: Firstly, existing coevolutionary frameworks for
with rough encoding by many-objective optimization hold a strong cooperation
between multiple populations, where the populations
⌈ ⌉ cooperate in mating selection, offspring generation, and
1 ∑
rear
y1 = x2i−1 + 1 environmental selection [56], [57]; by contrast, the pro-
rear − head + 1 posed framework holds a weak cooperation that only
i=head
y2 = head shares the offsprings generated by all the populations,
⌈ ⌉ where each population is evolved without the distur-
1 ∑
rear
, (8)
y3 = c − y1 − x2i + 1 bance of other populations and can better spread along
rear − head + 1 the whole Pareto front [58]. Secondly, existing coevolu-
i=head
y4 = rear − head + 1 tionary frameworks for large-scale optimization evolve
y5 = xd multiple populations with the same encoding scheme,
where each population corresponds to different part of
the decision variables [41], [44]; by contrast, the encoding
where d is the number of decision variables of each schemes of the populations are different in the proposed
aperture represented by fine encoding, c is the num- framework, which can better balance between exploita-
ber of columns in an aperture, and head and rear are tion and exploration.
the indexes of the first and last rows where the MLC There also exist some large-scale MOEAs using two
leaves are opened, respectively. As for a solution q with encoding schemes, where some solutions use the original
rough encoding, its decision variables y = (y1 , y2 , . . . , y5 ) encoding of a continuous MOP and some others use a
corresponding to each aperture are converted into x = shortened encoding to accelerate the convergence. For
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 8

Fine encoding Fine encoding


(Very small shape) (Repaired)
Algorithm 3: LocalSearchA(P1 , P2 )
× × × ××××× × × ×××××× Input: P1 (population with fine encoding), P2 (population
× × × ××××× × × ×××××× with rough encoding)
× × × ××××× × × ×××××× Output: P1 (repaired population with fine encoding), P2
× × × × ×× Repair
× × × × (repaired population with rough encoding)
× × × ×× × × × //Repair apertures in fine encoding
1 foreach solution p ∈ P1 do
× × × × ××× × × × ×× 2 foreach aperture x ∈ p do
× × × ××××× × × ×××××× 3 Repair x by (10);
× × × ××××× × × ××××××
Fine encoding Fine encoding //Repair apertures in rough encoding
(Disconnected shape) (Repaired) 4 foreach solution q ∈ P2 do
× × ××× ××× × × ××× ××× 5 foreach aperture y ∈ q do
× × × ××× × × × ××× 6 Repair y by (11);
× × ×× ××× Repair
× × × ××× 7 return P1 , P2 ;
× × ×× ×× × × ×× ××
× × ××× ××× × × × ×××
× × × ××× × × × ×××
× × ××× ××× × × ××× ××× each aperture are repaired by
× × ××× ××× × × ××× ××× x2i = min{x2i , c − x2i−1 }, 1 ≤ i ≤ r
{ ⌈ ⌉}
Rough encoding Rough encoding Smin − S
(Very small shape) (Repaired) x2i−1 = x2i−1 − max 0, ,
2(rear − head + 1)
× × × ×××× × × × × ××××× head ≤ i ≤ rear
× × × ×××× × × × × ××××× { ⌈ ⌉}
× × × ×××× × × × × ×× Smin − S , (10)
x2i = x2i − max 0, ,
× × × × × Repair
× × × ×× 2(rear − head + 1)
× × × × × × × × ×× head ≤ i ≤ rear
× × × ×××× × × × × ×× x2i−1 = min{x2i−1 , c − x2(i−1) − 1}, head < i ≤ rear
× × × ×××× × × × × ×××××
× × × ×××× × × × × ××××× x2i = min{x2i , c − x2(i−1)−1 − 1}, head < i ≤ rear

where c is the number of columns in an aperture, r is


Fig. 7. An illustrative example of the local search strategy for repairing
invalid solutions.
the number of rows in an aperture, Smin is the threshold
of small aperture, S is the number of opened grids in
the aperture, and head and rear are the indexes of the
example, the encoding is shortened by a weighted opti-
first and last rows where the MLC leaves are opened,
mization framework in [45], a linear search mechanism
respectively. In (10), the first formula ensures that each
in [59], and unsupervised neural networks in [47]. It
pair of MLC leaves do not intersect each other, the
should be noted that the idea of the proposed two
second and third formulas expand small apertures, and
encoding schemes is different from existing ones, as the
the last two formulas repair disconnected apertures. On
former is tailored for IMRT planning while the latter are
the other hand, for each solution q with rough encoding,
generic strategies for continuous black-box optimization.
its decision variables y = (y1 , y2 , . . . , y5 ) corresponding
Section IV-B will verify the superiority of the proposed
to each aperture are repaired by
algorithm by comparing it with existing coevolutionary
{ ⌈ ⌉}
algorithms and those with two encoding schemes. Smin − S
y1 = y1 − max 0, , rand ≤ 0.5
2y4
{ ⌈ ⌉}
Smin − S
y3 = y3 + max 0, , rand ≤ 0.5
y4
{ ⌈ ⌉} , (11)
C. Local Search Strategies of MOEA/IMRT Smin − S
y2 = y2 − max 0, , rand > 0.5
2y3
Although the bi-encoding coevolutionary framework { ⌈ ⌉}
Smin − S
is more suitable for IMRT planning than existing y4 = y4 + max 0, , rand > 0.5
y3
MOEAs, it is not efficient enough to find a set of diverse
plans. Therefore, two local search strategies are proposed where rand denotes a random number uniformly sam-
for a further enhancement of efficiency. As detailed in pled in [0, 1]. Since an aperture represented by rough
Fig. 7 and Algorithm 3, the first local search strategy encoding is always a connected rectangle, (11) only
LocalSearchA() is to ensure the clinical practicality of needs to expand small apertures along the vertical or
solutions, where all the aperture shapes in each solution horizontal direction with the same probability.
cannot be disconnected or too small (i.e., the number The second local search strategy LocalSearchB() aims
of opened grids is less than a predefined threshold). to enhance the population P1 with fine encoding by
To this end, for each solution p with fine encoding, its optimizing the beam weights. Represented by real vari-
decision variables x = (x1 , x2 , . . . , xd ) corresponding to ables, the beam weights are easier to optimize than the
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 9

TABLE I
D ETAILED I NFORMATION A BOUT S IX D ATASETS U SED IN THE E XPERIMENTS

Parameters Sneezer Renal RightLung Throat Cervical Esophagus


Number of M inDose() 5 2 2 2 3 1
Number of M axDose() 5 2 2 2 2 1
Planning tumor volume (PTV)
Number of P reDose() 1 2 2 2 2 1
Number of DV C() 5 2 2 2 3 1
Number of M axDose() 17 3 1 10 8 1
Organ at risk (OAR)
Number of DV C() 1 2 6 4 11 4
Numbers of beams 7 5 5 6 7 5
Numbers of apertures for each beam 6, 7, 5, 8, 7, 7, 8 9, 8, 12, 12, 9 10, 10, 9, 10, 11 8, 9, 8, 5, 8, 8 7, 6, 8, 6, 7, 8, 8 10, 11, 10, 10, 9
Number of rows of apertures for each beam 24, 24, 24, 24, 24, 24, 24 23, 23, 23, 23, 22 27, 28, 27, 27, 27 21, 21, 21, 21, 21, 21 30, 30, 30, 30, 30, 30, 30 21, 21, 21, 21, 21
Number of decision variables for rough coding 240 250 250 230 250 250
Number of decision variables for fine coding 2352 2332 2770 1978 3050 2150

Algorithm 4: LocalSearchB(P1 ) RVEA are effective MOEAs tailored for many-objective


Input: P1 (population with fine coding) optimization, where Two Arch2 has a coevolutionary
Output: P1 (optimized population with fine coding) framework and RVEA is based on objective decom-
1 PN ← Non-dominated solutions in P1 ; position. CCGDE3, WOF-SMPSO, and LMOCSO are
2 W ← Randomly generate a population with only beam large-scale MOEAs based on variable grouping, decision
weights;
space reduction, and novel search strategy, respectively;
3 Randomly select a solution from PN for each solution in
W , then evaluate W ; besides, WOF-SMPSO adopts two encoding schemes.
4 while termination criterion is not fulfilled do Cotrutz’s algorithm and Wang’s algorithm are direct
5 W ′ ← Select parents from W via binary tournament aperture optimization methods based on genetic algo-
selection; rithm and conjugate gradient, respectively, and Cao’s
6 O ← Generate offsprings based on W ′ ; algorithm is a two-step optimization method based on
7 Randomly select a solution from PN for each solution
in O, then evaluate O; conjugate gradient.
8 W ← W ∪ O;
9 W ← Truncate W via environmental selection; A. Parameter Settings
10 P1 ← P1 ∪ W ;
1) Datasets: The compared algorithms are tested on six
11 P1 ← Truncate P1 via environmental selection;
12 return P1 ; real-world tumor cases, including sneezer, renal, right
lung, throat, cervical, and esophagus cancers. For each
case, the medical physicist has defined the number of
aperture shapes, however, they interact with the aperture beams, the number of apertures for each beam, the size
shapes and may not be well optimized. Therefore, this of each aperture, and a number of objective functions
local search strategy merely optimizes all the beam with corresponding parameters for the regions in PTV
weights in the solutions. In fact, some existing IMRT and OAR. Detailed information about these datasets is
planning methods only optimize the beam weights for presented in Table I, where the number of real variables
simplicity [7], [26], [29]. As described in Algorithm 4, varies from 230 to 250 for rough encoding and varies
the optimization of beam weights employs the same from 1978 to 3050 for fine encoding.
selection strategies and genetic operators as the main 2) Parameters in Algorithms: The parameters in the
framework of MOEA/IMRT. Since a solution consisting compared algorithms are set as suggested in their origi-
of beam weights cannot be evaluated, each solution is nal papers. In RVEA, the rate of change of penalty is set
integrated with the aperture shapes of a non-dominated to 2 and the frequency of reference vector adaptation is
solution randomly selected from P1 . After the optimiza- set to 0.1. In CCGDE3, the number of groups is set to 2
tion of beam weights, the final population is combined and the size of each subpopulation is set to 40. In WOF-
with P1 and truncated by environmental selection. This SMPSO, the variables are divided into 4 groups by or-
local search strategy is performed with a population size dered grouping and aggregated by interval-intersection
of 20 and 2000 function evaluations after every 100 gen- transformation, the number of evaluations for original
erations of MOEA/IMRT, which does not consume much problem is set to 1000, the number of evaluations for
computational resource. Instead, it brings a significant transformed problem is set to 500, and the fraction
performance improvement as verified in Section IV-C. of function evaluations for transformed problem is set
to 0.5. In the proposed MOEA/IMRT, the number of
IV. E XPERIMENTAL S TUDIES exchanged solutions K is set to 10. As for the reproduc-
The proposed MOEA/IMRT is compared with eight tion operators, Two Arch2, RVEA, Cotrutz’s algorithm,
representative algorithms in the experiments, including and MOEA/IMRT use simulated binary crossover [52]
Two Arch2 [57], RVEA [60], CCGDE3 [41], WOF-SMPSO and polynomial mutation [53], where the probability of
[45], LMOCSO [48], Cotrutz’s algorithm [61], Cao’s algo- crossover is set to 1, the probability of mutation is set
rithm [62], and Wang’s algorithm [14]. Two Arch2 and to 1/D (D is the number of decision variables), and the
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 10

1 1 1
Dose-volume constraint
Relative volume (%)

Relative volume (%)

Relative volume (%)


0.8 Cumulative distribution 0.8 Maximum dose 0.8 (d , r ) (d ,r )
dvc1 1 dvc2 2
of the doses in PTV dmax
0.6 0.6 0.6
Maximum dose
0.4 d 0.4 0.4
max
Cumulative distribution Cumulative distribution
0.2 Minimum dose 0.2 of the doses in OAR 0.2 of the doses in OAR
dmin
0 0 0
0 1000 2000 3000 4000 0 1000 2000 3000 4000 5000 6000 0 1000 2000 3000 4000
Dose (/cGy) Dose (/cGy) Dose (/cGy)

(a) Maximum and minimum doses of PTV. (b) Maximum dose of OAR. (Ideal case: the (c) Dose-volume constraint of OAR. (Ideal
(Ideal case: the vertical part of the curve is curve is to the left of the vertical line) case: the curve is to the lower left of the
between the two vertical lines) two points)

Fig. 8. Illustration of different objectives in a dose volume histogram (DVH). A point (x, y) on the curve indicates that the ratio of doses greater
than x in the PTV or OAR is y.

distribution index is set to 20; CCGDE3 uses differential are reported. Besides, the Friedman test with Bonferroni
evolution and polynomial mutation, where both the correction [63] at a significance level of 0.05 is adopted
learning rate F and the crossover rate CR are set to for statistical test, where ‘+’, ‘−’, and ‘≈’ indicate that
0.5; WOF-SMPSO uses particle swarm optimization and the HV values obtained by an algorithm are significantly
polynomial mutation; LMOCSO uses competitive swarm better, significantly worse, and statistically similar to the
optimizer and polynomial mutation; Cao’s algorithm proposed MOEA/IMRT, respectively.
and Wang’s algorithm uses conjugate gradient method. 6) Dose-Volume Histogram (DVH): The DVH is also
3) Problem-Dependent Strategies: In spite of the differ- used to visually assess the quality of plans, which is the
ent numbers of objective functions in the datasets, they main assessment criterion used by medical physicists. A
are aggregated into five objectives by (7) and optimized DVH depicts the cumulative distribution of the doses in
by all the compared MOEAs. As for Cotrutz’s algorithm, PTV and OAR, where a point (x, y) on a curve indicates
Cao’s algorithm, and Wang’s algorithm, they aggregate that the ratio of doses greater than x in the PTV or
all the objective functions into a single one. Except OAR is y. Fig. 8 illustrates a cumulative distribution
for the proposed MOEA/IMRT adopting two encoding curve together with different objectives in a DVH, where
schemes, all the other compared algorithms represent the vertical part of the curve should be between the
solutions with fine encoding. To ensure the validity of two vertical lines of maximum and minimum doses in
the solutions, all the compared MOEAs perform the Fig. 8(a), the curve should be to the left of the vertical line
LocalSearchA() described in Algorithm 3 once an off- of maximum dose in Fig. 8(b), and the curve should be
spring is generated. Since Cotrutz’s algorithm, Cao’s to the lower left of the points of dose-volume constraint
algorithm, and Wang’s algorithm can only generate one in Fig. 8(c). If the curve does not meet any of the above
solution at a time, a set of uniformly distributed weight conditions, it implies that the plan is not optimal on the
vectors are sampled for the aggregation of objective corresponding objective.
functions, where one solution is obtained by optimizing
the single objective aggregated by each weight vector
and all the solutions constitute the final population. B. Comparisons Between MOEA/IMRT and State-of-the-Art
4) Population Size and Termination Criterion: For fair- Table II lists the HV results obtained by the nine
ness, the population size of all the compared MOEAs is compared algorithms on six datasets. It is clear that the
set to 100 and the maximum number of function evalu- proposed MOEA/IMRT obtains the best performance on
ations of all the compared MOEAs is set to 500000. Be- all the datasets, whose HV values are always signifi-
sides, Cotrutz’s algorithm, Cao’s algorithm, and Wang’s cantly better than all the other compared algorithms.
algorithm are performed 100 times and the maximum For a visual comparison, Fig. 9 compares the objective
number of function evaluations is set to 5000 each time. values of the populations with median HV obtained
5) Performance Indicator: Since the true Pareto fronts of by the proposed MOEA/IMRT and each of the other
real-world problems are unknown, the HV indicator [54] algorithms on the Esophagus dataset. It can be observed
is adopted to assess the performance of the compared that the population obtained by MOEA/IMRT has bet-
algorithms. To determine the reference point for HV ter convergence than those obtained by Two Arch2,
calculation, the non-dominated solutions obtained by CCGDE3, WOF-SMPSO, Cotrutz’s algorithm, Cao’s al-
all the algorithms in all the runs are collected, and the gorithm, and Wang’s algorithm, and has better diver-
reference point consists of the maximum objective values sity than those obtained by RVEA and LMOCSO. For
of the non-dominated solutions on all dimensions. To Two Arch2 and RVEA tailored for many-objective opti-
eliminate the influence of randomness, each algorithm mization, the convergence performance of RVEA is bet-
is performed on each dataset for 30 independent runs, ter than Two Arch2, which means that the coevolution-
where the mean and standard deviation of the HV values ary framework of Two Arch2 is ineffective for the many
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 11

TABLE II
HV R ESULTS O BTAINED BY N INE A LGORITHMS ON S IX D ATASETS OF IMRT P LANNING . T HE B EST R ESULT IN E ACH R OW IS H IGHLIGHTED

Problem Two Arch2 RVEA CCGDE3 WOF-SMPSO LMOCSO Cotrutz’s algorithm Cao’s algorithm Wang’s algorithm MOEA/IMRT

Esophagus 9.6064e-1 (8.85e-3) − 9.6384e-1 (3.75e-3) − 0.0000e+0 (0.00e+0) − 9.5540e-1 (9.52e-3) − 9.5822e-1 (1.31e-2) − 3.4490e-2 (3.22e-2) − 0.0000e+0 (0.00e+0) − 9.1906e-1 (1.00e-2) − 9.8045e-1 (2.61e-3)

Renal 9.9330e-1 (1.72e-3) − 9.8747e-1 (1.95e-3) − 0.0000e+0 (0.00e+0) − 9.7799e-1 (8.29e-3) − 9.8099e-1 (1.46e-3) − 9.0421e-1 (2.20e-2) − 0.0000e+0 (0.00e+0) − 9.8749e-1 (6.67e-4) − 9.9768e-1 (3.60e-4)

RightLung 8.8601e-1 (1.13e-2) − 7.5203e-1 (2.43e-1) − 2.4031e-2 (5.33e-2) − 7.2896e-1 (2.64e-1) − 8.4637e-1 (1.72e-2) − 4.8269e-1 (2.72e-1) − 0.0000e+0 (0.00e+0) − 7.2737e-1 (4.29e-3) − 9.0784e-1 (6.61e-3)

Sneezer 9.8459e-1 (1.41e-3) − 9.7737e-1 (1.89e-3) − 0.0000e+0 (0.00e+0) − 8.9300e-1 (1.27e-1) − 9.8122e-1 (1.06e-3) − 9.6587e-1 (5.76e-3) − 1.0271e-1 (1.41e-2) − 9.7198e-1 (2.43e-3) − 9.8680e-1 (1.10e-3)

Throat 8.4575e-1 (1.77e-2) − 5.9789e-1 (3.27e-2) − 0.0000e+0 (0.00e+0) − 5.8518e-1 (1.18e-1) − 2.2960e-1 (2.14e-1) − 0.0000e+0 (0.00e+0) − 0.0000e+0 (0.00e+0) − 0.0000e+0 (0.00e+0) − 8.5912e-1 (9.62e-3)

Cervical 9.6661e-1 (3.46e-3) − 9.6897e-1 (5.40e-3) − 1.1003e-1 (5.95e-2) − 9.2699e-1 (3.31e-2) − 9.7579e-1 (1.67e-3) − 8.4165e-1 (1.33e-2) − 0.0000e+0 (0.00e+0) − 9.4716e-1 (3.58e-3) − 9.8394e-1 (1.61e-3)

+/ − / ≈ 0/6/0 0/6/0 0/6/0 0/6/0 0/6/0 0/6/0 0/6/0 0/6/0

Fig. 9. Parallel coordinates of the objective values of the populations with median HV obtained by the compared algorithms on the Esophagus
dataset. MOEA/IMRT has better convergence performance than Two Arch2, CCGDE3, WOF-SMPSO, Cotrutz’s algorithm, Cao’s algorithm, and
Wang’s algorithm, and better diversity performance than RVEA and LMOCSO.

objectives of IMRT; besides, the reference vector based and OAR2, it can be found that all the algorithms can
selection strategy of RVEA is imperfect since it has worse almost meet the requirement of dose-volume constraints.
diversity performance than MOEA/IMRT. For CCGDE3, For the curves of OAR3, all the algorithms can mostly
WOF-SMPSO, and LMOCSO tailored for large-scale op- meet the requirement of maximum dose. As for the most
timization, the novel search strategy of LMOCSO is important curves of PTV, only RVEA, LMOCSO, and
more effective than the coevolutionary framework and the proposed MOEA/IMRT can meet the requirement of
variable grouping strategy of CCGDE3 and the two both maximum and minimum doses. The band of PTV
encoding schemes and decision space reduction strategy obtained by MOEA/IMRT between the vertical lines of
of WOF-SMPSO, since the former exhibits much better maximum and minimum doses is obviously wider than
convergence performance; besides, the diversity perfor- those obtained by RVEA and LMOCSO, which indi-
mance of LMOCSO is also worse than MOEA/IMRT. cates a better diversity performance of MOEA/IMRT.
For Cotrutz’s algorithm, Cao’s algorithm, and Wang’s In addition, Figs. S1–S5 in the Supplementary Materials
algorithm tailored for IMRT planning, they have worse show the DVH of the populations obtained on the other
convergence performance than MOEA/IMRT since the five datasets, where MOEA/IMRT can also obtain a set
aperture shapes cannot be well optimized by using ge- of well-converged and diversified solutions on all the
netic algorithm (without local search), MLC sequencing, datasets.
and enumeration, respectively.
Moreover, Fig. 10 shows the DVH of the populations
with median HV obtained by the compared algorithms C. Verification of the Core Components in MOEA/IMRT
on the Esophagus dataset. To better reflect the popula- Firstly, MOEA/IMRT is compared to a variant without
tion diversity, the range of all the curves in a population the local search strategy LocalSearchB() to verify the
rather than a single curve is plotted, where a curve effectiveness of the local search strategy for enhancing
corresponds to the cumulative distribution of the doses the population with fine encoding. According to the
in PTV or OAR of one solution. For the curves of OAR1 convergence profiles of mean HV shown in Fig. 11,
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 12

Fig. 10. DVH of the populations with median HV obtained by the compared algorithms on the Esophagus dataset. The meanings of the elements
in the figures are referred to Fig. 8.

0.95
of the solutions obtained by MOEA/IMRT with and
without local search, where the curves corresponding
0.9
to the solution with the smallest sum of all objectives
0.85 in the population with median HV are shown. It can
0.8
be found that the two algorithms obtain the same good
With LocalSearch
Without LocalSearch curves of OAR1–OAR3, while the algorithm without
1 2 3 4 5 local search obtains a curve of PTV obviously exceeding
Number of evaluations 105
1 the maximum dose. Therefore, performing local search
With local search (PTV) Without local search (OAR2)
0.9 With local search (OAR1)
With local search (OAR2)
Without local search (OAR3)
Maximum/Minimum dose (PTV)
on the population with fine encoding can improve the
0.8 With local search (OAR3)
Without local search (PTV)
Dose-volume cons (OAR1)
Dose-volume cons (OAR2)
performance of MOEA/IMRT indeed.
0.7 Without local search (OAR1) Maximum dose (OAR3)
Then, MOEA/IMRT is compared to three variants
Relative volume (%)

0.6

0.5
with different values of K, i.e., the number of exchanged
0.4
solutions in Algorithm 2. According to the results shown
0.3
in Fig. 12, the algorithms with different values of K
0.2 gain very similar HV values, which means that the
0.1 performance of MOEA/IMRT is not sensitive to the
0
0 1000 2000 3000 4000 5000 6000 7000
setting of K. Besides, K = 10 leads to a relatively
Dose (/cGy) better performance than the other settings, hence K is
suggested to be 10 in all cases.
Fig. 11. Convergence profiles of mean HV and DVH of the solutions
obtained by MOEA/IMRT with and without local search on the Lastly, MOEA/IMRT is compared to two variants,
Esophagus dataset.
where one variant evolves only a population with rough
encoding and the other variant evolves only a popula-
the elimination of the local search strategy decreases tion with fine encoding. According to the HV results
the convergence speed of MOEA/IMRT and leads to a obtained on all the datasets given in Table SI in the
worse HV value. Besides, the figure also shows the DVH Supplementary Materials, the MOEA/IMRT with two
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 13

as involving the angles of beams [64] and the numbers


0.9

0.8
of apertures in the decision variables [11].
0.7

0.6
R EFERENCES
0.5

0.4 [1] B. W. Fischer-Valuck, Y. J. Rao, and J. M. Michalski, “Intensity-


0.3 modulated radiotherapy for prostate cancer,” Translational Androl-
0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Number of evaluations 105
ogy and Urology, vol. 7, no. 3, pp. 297–307, 2018.
[2] R. Bokrantz, “Multicriteria optimization for managing tradeoffs
in radiation therapy treatment planning,” Ph.D. dissertation, KTH
Royal Institute of Technology, 2013.
[3] A. N. Hasto and W. Rena, “Optimization of time-dose fractiona-
tion radiotherapy scheme by simulated annealing with consider-
ation of biological factor,” Journal of Physics: Conference Series, vol.
1090, p. 012040, 2018.
[4] A. Brahme, “Optimization of stationary and moving beam radia-
tion therapy techniques,” Radiotherapy and Oncology, vol. 12, no. 2,
pp. 129–140, 1988.
[5] B. K. Lind, “Properties of an algorithm for solving the inverse
problem in radiation therapy,” Inverse Problems, vol. 6, no. 3, pp.
415–426, 1990.
[6] A. T. Redpath, B. L. Vickery, and D. H. Wright, “A new technique
for radiotherapy planning using quadratic programming,” Physics
Fig. 12. Convergence profiles of mean HV and DVH of the populations in Medicine & Biology, vol. 21, no. 5, p. 781, 1976.
with median HV obtained by MOEA/IMRT with different values of [7] S. V. Spirou and C. S. Chui, “A gradient inverse planning al-
K on the Esophagus dataset. gorithm with dose-volume constraints,” Medical Physics, vol. 25,
no. 3, pp. 321–333, 1998.
[8] S. Webb, “Optimization by simulated annealing of three-
encoding schemes obtains obviously better overall per- dimensional conformal treatment planning for radiation fields
defined by a multileaf collimator,” Physics in Medicine & Biology,
formance than those with only rough encoding or fine vol. 36, no. 9, pp. 1201–1226, 1991.
encoding. Hence, the effectiveness of the two encoding [9] G. A. Ezzell, “Genetic and geometric optimization of three-
schemes can be verified. dimensional radiation therapy treatment planning,” Medical
Physics, vol. 23, no. 3, pp. 293–305, 1996.
[10] S. P. Goldman, J. Z. Chen, and J. J. Battista, “Feasibility of a
V. C ONCLUSIONS AND F UTURE W ORK fast inverse dose optimization algorithm for IMRT via matrix
inversion without negative beamlet intensities,” Medical Physics,
In order to improve the efficiency of IMRT planning, a vol. 32, no. 9, pp. 3007–3016, 2005.
set of diverse plans are expected to be obtained in a sin- [11] M. Ehrgott, Ç. Güler, H. W. Hamacher, and L. Shao, “Mathemati-
cal optimization in intensity modulated radiation therapy,” Annals
gle optimization process. However, most existing IMRT of Operations Research, vol. 175, pp. 309–365, 2010.
planning methods can only obtain a single solution at a [12] A. Cassioli and J. Unkelbach, “Aperture shape optimization for
time, and existing MOEAs can hardly be employed due IMRT treatment planning,” Physics in Medicine & Biology, vol. 58,
no. 2, pp. 301–318, 2012.
to their poor performance on large-scale combinatorial [13] D. M. Shepard, M. A. Earl, X. A. Li, S. Naqvi, and C. Yu,
many-objective optimization problems. Therefore, this “Direct aperture optimization: A turnkey solution for step-and-
paper has proposed a new MOEA for IMRT planning, shoot IMRT,” Medical Physics, vol. 29, no. 6, pp. 1007–1018, 2002.
[14] J. Wang, X. Pei, R. Cao, L. Hu, and Y. Wu, “A fast direct aperture
which tackles the high-dimensional discrete search space optimization method for intensity modulated radiation therapy,”
of IMRT planning by suggesting a bi-encoding coevolu- Chinese Journal of Medical Physics, vol. 32, no. 1, pp. 4–7, 2015.
tionary framework and two local search strategies. Ex- [15] H. E. Romeijn, R. K. Ahuja, J. F. Dempsey, and A. Kumar, “A
column generation approach to radiation therapy treatment plan-
perimental results have demonstrated that the use of two ning using aperture modulation,” SIAM Journal on Optimization,
encoding schemes can well balance between exploitation vol. 15, no. 3, pp. 838–862, 2012.
and exploration, and the local search strategies can [16] E. Salari and J. Unkelbach, “A column-generation-based method
for multi-criteria direct aperture optimization,” Physics in Medicine
further improve the convergence speed. In comparison & Biology, vol. 58, no. 3, pp. 621–639, 2013.
to state-of-the-art algorithms, the proposed algorithm [17] R. Cao, X. Pei, H. Zheng, L. Hu, and Y. Wu, “Direct aperture
exhibits better convergence and diversity performance optimization based on genetic algorithm and conjugate gradient
on various clinical cases. in intensity modulated radiation therapy,” Chinese Medical Journal,
vol. 127, no. 23, pp. 4152–4153, 2014.
Mathematical programming methods have fast con- [18] J. Yang, P. Zhang, L. Zhang, and Z. Gui, “A gradient-based direct
vergence speed but can obtain only one solution in a aperture optimization,” Journal of Biomedical Engineering, vol. 35,
single run, while MOEAs have relatively slow conver- no. 3, pp. 358–367, 2018.
[19] L. T. Oanh, D. T. Tai, T. T. H. Loan, T. H. Minh, T. V. Minh, and
gence speed but can obtain multiple solutions simultane- J. C. L. Chow, “Dosimetric evaluation of lung treatment plans
ously. Although the proposed MOEA has demonstrated produced by the Prowess Panther system using Monte Carlo
high effectiveness in IMRT planning, it is reasonable to simulation,” Biomedical Physics & Engineering Express, vol. 5, p.
055005, 2019.
take full advantage of both mathematical programming [20] S. Breedveld, D. Craft, R. van Haveren, and B. Heijmen, “Multi-
methods and MOEAs for achieving better convergence criteria optimization and decision-making in radiotherapy,” Euro-
and diversity performance in the future. In addition, it is pean Journal of Operational Research, vol. 277, no. 1, pp. 1–19, 2019.
[21] D. Craft and T. Bortfeld, “How many plans are needed in an IMRT
desirable to build more complex optimization models to multi-objective plan database?” Physics in Medicine & Biology,
further reduce the requirement of human expertise, such vol. 53, p. 2785, 2008.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 14

[22] D. Craft and C. Richter, “Deliverable navigation for multicriteria [42] L. M. Antonio, C. A. Coello Coello, M. A. R. Morales, S. G. Bram-
step and shoot IMRT treatment planning,” Physics in Medicine & bila, J. F. González, and G. C. Tapia, “Coevolutionary operations
Biology, vol. 58, p. 87, 2013. for large scale multi-objective optimization,” in Proceedings of the
[23] W. Hong, P. Yang, and K. Tang, “Evolutionary computation for 2020 IEEE Congress on Evolutionary Computation, 2020.
large-scale multi-objective optimization: A decade of progresses,” [43] M. Li and J. Wei, “A cooperative co-evolutionary algorithm for
International Journal of Automation and Computing, vol. 18, pp. 155– large-scale multi-objective optimization problems,” in Proceedings
169, 2021. of the 2018 Annual Conference on Genetic and Evolutionary Compu-
[24] Y. Tian, L. Si, X. Zhang, R. Cheng, C. He, K. C. Tan, and Y. Jin, tation Conference, 2018, pp. 1716–1721.
“Evolutionary large-scale multi-objective optimization: A survey,” [44] X. Zhang, Y. Tian, R. Cheng, and Y. Jin, “A decision variable
ACM Computing Surveys, vol. 54, no. 8, p. 174, 2022. clustering-based evolutionary algorithm for large-scale many-
[25] M. Lahanas, D. Baltas, and N. Zamboglou, “A hybrid evolu- objective optimization,” IEEE Transactions on Evolutionary Compu-
tionary algorithm for multi-objective anatomy-based dose opti- tation, vol. 22, no. 1, pp. 97–112, 2018.
mization in high-dose-rate brachytherapy,” Physics in Medicine & [45] H. Zille, H. Ishibuchi, S. Mostaghim, and Y. Nojima, “A frame-
Biology, vol. 48, no. 3, pp. 399–415, 2003. work for large-scale multiobjective optimization based on prob-
[26] Y. Chen, D. Michalski, C. Houser, and J. M. Galvin, “A determinis- lem transformation,” IEEE Transactions on Evolutionary Computa-
tic iterative least-squares algorithm for beam weight optimization tion, vol. 22, no. 2, pp. 260–275, 2018.
in conformal radiotherapy,” Physics in Medicine & Biology, vol. 47, [46] C. He, L. Li, Y. Tian, X. Zhang, R. Cheng, Y. Jin, and X. Yao,
no. 10, pp. 1647–1658, 2002. “Accelerating large-scale multiobjective optimization via problem
[27] R. Cao, Y. Wu, X. Pei, J. Jing, G. Li, M. Cheng, G. Li, and L. Hu, reformulation,” IEEE Transactions on Evolutionary Computation,
“Multi-objective optimization of inverse planning for accurate vol. 23, no. 6, pp. 949–961, 2019.
radiotherapy,” Chinese Physics C, vol. 35, no. 3, pp. 313–317, 2011. [47] Y. Tian, C. Lu, X. Zhang, K. C. Tan, and Y. Jin, “Solving large-
[28] J. Yang, P. Zhang, L. Zhang, H. Shu, B. Li, and Z. Gui, “Parti- scale multiobjective optimization problems with sparse optimal
cle swarm optimizer for weighting factor selection in intensity- solutions via unsupervised neural networks,” IEEE Transactions
modulated radiation therapy optimization algorithms,” Physica on Cybernetics, vol. 51, no. 6, pp. 3115–3128, 2021.
Medica, vol. 33, pp. 136–145, 2017. [48] Y. Tian, X. Zheng, X. Zhang, and Y. Jin, “Efficient large-scale multi-
[29] C. Cotrutz, M. Lahanas, C. Kappas, and D. Baltas, “A multiob- objective optimization based on a competitive swarm optimizer,”
jective gradient-based dose optimization algorithm for external IEEE Transactions on Cybernetics, vol. 50, no. 8, pp. 3696–3708, 2020.
beam conformal radiotherapy,” Physics in Medicine & Biology, [49] W. Hong, K. Tang, A. Zhou, H. Ishibuchi, and X. Yao, “A scalable
vol. 46, no. 8, pp. 2161–2175, 2001. indicator-based evolutionary algorithm for large-scale multiobjec-
[30] G. Li and D. Cao, “A multi-objective particle swarm algorithm tive optimization,” IEEE Transactions on Evolutionary Computation,
for the optimization of IMRT inverse planning,” in Proceedings vol. 23, no. 3, pp. 525–537, 2018.
of the 3rd International Conference on Biomedical Engineering and [50] C. He, S. Huang, R. Cheng, K. C. Tan, and Y. Jin, “Evolutionary
Informatics, 2010. multiobjective optimization driven by generative adversarial net-
[31] A. Jalalimanesh, H. S. Haghighi, A. Ahmadi, H. Hejazian, and works (GANs),” IEEE Transactions on Cybernetics, vol. 51, no. 6,
M. Soltani, “Multi-objective optimization of radiotherapy: dis- pp. 3129–3142, 2021.
tributed Q-learning and agent-based simulation,” Journal of Ex-
[51] H. Zille and S. Mostaghim, “Comparison study of large-scale
perimental & Theoretical Artificial Intelligence, vol. 29, no. 5, pp.
optimisation techniques on the LSMOP benchmark functions,”
1071–1086, 2016.
in Proceedings of the 2017 IEEE Symposium Series on Computational
[32] T. Spalke, D. Craft, and T. Bortfeld, “Analyzing the main
Intelligence, 2017.
trade-offs in multiobjective radiation therapy treatment planning
[52] K. Deb and R. B. Agrawal, “Simulated binary crossover for
databases,” Physics in Medicine & Biology, vol. 54, no. 12, p. 3741,
continuous search space,” Complex Systems, vol. 9, no. 4, pp. 115–
2009.
148, 1995.
[33] R. Bokrantz, “Distributed approximation of Pareto surfaces in
multicriteria radiation therapy treatment planning,” Physics in [53] K. Deb and M. Goyal, “A combined genetic adaptive search
Medicine & Biology, vol. 58, no. 11, p. 3501, 2013. (GeneAS) for engineering design,” vol. 26, no. 4, 1996, pp. 30–
[34] S. Breedveld, P. R. M. Storchi, and B. J. M. Heijmen, “The 45.
equivalence of multi-criteria methods for radiotherapy plan opti- [54] L. While, P. Hingston, L. Barone, and S. Huband, “A faster
mization,” Physics in Medicine & Biology, vol. 54, no. 23, p. 7199, algorithm for calculating hypervolume,” IEEE Transactions on
2009. Evolutionary Computation, vol. 10, no. 1, pp. 29–38, 2006.
[35] R. van Haveren, S. Breedveld, M. Keijzer, P. Voet, B. Heijmen, [55] E. Zitzler, M. Laumanns, and L. Thiele, “SPEA2: Improving the
and W. Ogryczak, “Lexicographic extension of the reference point strength Pareto evolutionary algorithm for multiobjective opti-
method applied in radiation therapy treatment planning,” Euro- mization,” in Proceedings of the Fifth Conference on Evolutionary
pean Journal of Operational Research, vol. 263, no. 1, pp. 247–257, Methods for Design, Optimization and Control with Applications to
2017. Industrial Problems, 2001, pp. 95–100.
[36] K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and eli- [56] R. Wang, R. C. Purshouse, and P. J. Fleming, “Preference-
tist multiobjective genetic algorithm: NSGA-II,” IEEE Transactions inspired coevolutionary algorithms for many-objective optimiza-
on Evolutionary Computation, vol. 6, no. 2, pp. 182–197, 2002. tion,” IEEE Transactions on Evolutionary Computation, vol. 17, no. 4,
[37] Y. Tian, R. Cheng, X. Zhang, Y. Su, and Y. Jin, “A strengthened pp. 474–494, 2013.
dominance relation considering convergence and diversity for [57] H. Wang, L. Jiao, and X. Yao, “Two Arch2: An improved two-
evolutionary many-objective optimization,” IEEE Transactions on archive algorithm for many-objective optimization,” IEEE Trans-
Evolutionary Computation, vol. 23, no. 2, pp. 331–345, 2019. actions on Evolutionary Computation, vol. 19, no. 4, pp. 524–541,
[38] M. Li, S. Yang, and X. Liu, “Shift-based density estimation for 2015.
Pareto-based algorithms in many-objective optimization,” IEEE [58] Y. Tian, T. Zhang, J. Xiao, X. Zhang, and Y. Jin, “A coevolu-
Transactions on Evolutionary Computation, vol. 18, no. 3, pp. 348– tionary framework for constrained multi-objective optimization
365, 2014. problems,” IEEE Transactions on Evolutionary Computation, vol. 25,
[39] K. Li, K. Deb, Q. Zhang, and S. Kwong, “Combining dominance no. 1, pp. 102–116, 2021.
and decomposition in evolutionary many-objective optimization,” [59] H. Zille and S. Mostaghim, “Linear search mechanism for multi-
IEEE Transactions on Evolutionary Computation, vol. 19, no. 5, pp. and many-objective optimisation,” in Proceedings of the 2019 In-
694–716, 2015. ternational Conference on Evolutionary Multi-Criterion Optimization,
[40] Y. Tian, R. Cheng, X. Zhang, F. Cheng, and Y. Jin, “An indicator 2019, pp. 399–410.
based multiobjective evolutionary algorithm with reference point [60] R. Cheng, Y. Jin, M. Olhofer, and B. Sendhoff, “A reference vector
adaptation for better versatility,” IEEE Transactions on Evolutionary guided evolutionary algorithm for many-objective optimization,”
Computation, vol. 22, no. 4, pp. 609–622, 2018. IEEE Transactions on Evolutionary Computation, vol. 20, no. 5, pp.
[41] L. M. Antonio and C. A. Coello Coello, “Use of cooperative 773–791, 2016.
coevolution for solving large scale multiobjective optimization [61] C. Cotrutz and L. Xing, “Segment-based dose optimization using
problems,” in Proceedings of the 2013 IEEE Congress on Evolutionary a genetic algorithm,” Physics in Medicine & Biology, vol. 48, no. 18,
Computation, 2013, pp. 2758–2765. pp. 2987–2998, 2003.
IEEE TRANSACTIONS ON EVOLUTIONARY COMPUTATION, VOL. , NO. , MONTH YEAR 15

[62] R. Cao, X. Pei, H. Zheng, J. Jing, M. Cheng, G. Li, and Y. Wu, Xingyi Zhang (SM’18) received the B.Sc. degree
“Beam intensity map optimization based on conjugate gradient in from Fuyang Normal College, Fuyang, China,
intensity modulated radiation treatment,” Nuclear Physics Review, in 2003, and the M.Sc. and Ph.D. degrees from
vol. 27, no. 2, pp. 212–217, 2010. Huazhong University of Science and Technol-
[63] J. Derrac, S. Garcia, D. Molina, and F. Herrera, “A practical tutorial ogy, Wuhan, China, in 2006 and 2009, respec-
on the use of nonparametric statistical tests as a methodology tively.
for comparing evolutionary and swarm intelligence algorithms,” He is currently a Professor with the School
Swarm and Evolutionary Computation, vol. 1, no. 1, pp. 3–18, 2011. of Artificial Intelligence, Anhui University,
[64] S. Breedveld, P. Storchi, P. Voet, and B. Heijmen, “iCycle: Inte- Hefei, China. His current research interests in-
grated, multicriterial beam angle, and profile optimization for clude unconventional models and algorithms of
generation of coplanar and noncoplanar IMRT plans,” Medical computation, evolutionary multi-objective op-
Physics, vol. 39, no. 2, pp. 951–963, 2012. timization, and logistic scheduling. He is the recipient of the 2018
and 2021 IEEE Transactions on Evolutionary Computation Outstanding
Paper Award and the 2020 IEEE Computational Intelligence Magazine
Outstanding Paper Award.
Xi Pei received the B.Sc. degree from Nan-
jing University of Aeronautics and Astronau-
Ye Tian received the B.Sc., M.Sc., and Ph.D.
tics, Nanjing, China, in 2006, and the Ph.D.
degrees from Anhui University, Hefei, China,
degrees from University of Chinese Academy
in 2012, 2015, and 2018, respectively.
of Sciences, Hefei, China, in 2012.
He is currently an Associate Professor with
He is currently an Associate Professor with
the Institutes of Physical Science and Infor-
the School of Nuclear Science and Technol-
mation Technology, Anhui University, Hefei,
ogy, University of Science and Technology of
China. His current research interests include
China, Hefei, China. His main research interests
evolutionary computation and its applications.
include key technologies of radiotherapy and
He is the recipient of the 2018 and 2021
their applications.
IEEE Transactions on Evolutionary Computa-
tion Outstanding Paper Award, the 2020 IEEE Kay Chen Tan (SM’08-F’14) received the B.Eng.
Computational Intelligence Magazine Outstanding Paper Award, and (First Class Hons.) degree in electronics and
the 2022 IEEE Computational Intelligence Society Outstanding PhD electrical engineering and the Ph.D. degree
Dissertation Award. from the University of Glasgow, Glasgow, U.K.
He is currently a Chair Professor of the Depart-
ment of Computing at the Hong Kong Poly-
technic University, Hong Kong. He has pub-
lished over 200 journal papers and six books,
and holds one U.S. patent on surface defect
Yuandong Feng received the B.Sc. degree from detection.
Hunan Normal University, Changsha, China, Prof. Tan is currently the Vice-President (Pub-
in 2018, where he is currently pursuing the lications) of IEEE Computational Intelligence Society, USA. He has
M.Sc. degree with the School of Computer Sci- served as the Editor-in-Chief of IEEE Transactions on Evolutionary
ence and Technology, Anhui University, Hefei, Computation from 2015-2020 and IEEE Computational Intelligence
China. Magazine from 2010-2013, and currently serves as the Editorial Board
His current research interests include large Member of over 10 journals. He was an IEEE Distinguished Lecturer
scale multi-objective optimization and its appli- Program (DLP) speaker and currently serves as the Chief Co-Editor of
cations. Springer Book Series on Machine Learning: Foundations, Methodolo-
gies, and Applications.
Yaochu Jin (SM’02-F’16) received the B.Sc.,
M.Sc., and Ph.D. degrees from Zhejiang Univer-
sity, Hangzhou, China, in 1988, 1991, and 1996,
respectively, and the Dr.-Ing. degree from Ruhr
Chao Wang received the B.Sc. degree from University Bochum, Germany, in 2001.
Suihua College, Suihua, China, in 2012, and the He is an Alexander von Humboldt Profes-
M.Sc. and Ph.D. degrees from Harbin Engineer- sor for Artificial Intelligence, Chair of Nature
ing University, Harbin, China, in 2015 and 2018, Inspired Computing and Engineering, Faculty
respectively. of Technology, Bielefeld University, Germany.
He is currently a Lecturer with the School of He is also a Distinguished Chair, Professor
Artificial Intelligence, Anhui University, Hefei, in Computational Intelligence, Department of
China. His main research interests include evo- Computer Science, University of Surrey, Guildford, U.K. He was a
lutionary computation, combinatorial optimiza- “Finland Distinguished Professor” of University of Jyväskylä, Fin-
tion, and intelligent transportation. land, “Changjiang Distinguished Visiting Professor”, Northeastern
University, China, and “Distinguished Visiting Scholar”, University
of Technology Sydney, Australia. His main research interests include
evolutionary optimization, evolutionary learning, trustworthy machine
learning, and evolutionary developmental systems.
Prof. Jin is presently the Editor-in-Chief of Complex & Intelligent
Systems. He was the Editor-in-Chief of the IEEE TRANSACTIONS
Ruifen Cao received the B.Sc. degree from ON COGNITIVE AND DEVELOPMENTAL SYSTEMS in 2016-2021
Henan University in 2004, and the Ph.D. de- and an IEEE Distinguished Lecturer in 2013–2015 and 2017–2019, the
gree from Hefei Institutes of Physical Science, Vice President for Technical Activities of the IEEE Computational
Chinese Academy of Sciences, Hefei, China, in Intelligence Society (2015–2016). He is the recipient of the 2018 and 2021
2010. IEEE Transactions on Evolutionary Computation Outstanding Paper
From 2010 to 2015, she was a Research Assis- Award, and the 2015, 2017, and 2020 IEEE Computational Intelligence
tant and Associate Researcher with the infor- Magazine Outstanding Paper Award. He was named by the Web of
mation science and technology. Since 2019, she Science as a Highly Cited Researcher from 2019 to 2021 consecutively.
has been an Associate Professor with the School He is a Member of Academia Europaea.
of Computer Science and Technology, Anhui
University. Her current research interest focuses
on medical image processing, artificial intelligence and its application
in radiotherapy.
View publication stats

You might also like