A hierarchical evolutionary algorithm for multiobjective optimization in IMRT

Deep reinforcement learning in radiation therapy planning optimization: A comprehensive review

PURPOSE

The formulation and optimization of radiation therapy plans are complex and time-consuming processes that heavily rely on the expertise of medical physicists. Consequently, there is an urgent need for automated optimization methods. Recent advancements in reinforcement learning, particularly deep reinforcement learning (DRL), show great promise for automating radiotherapy planning. This review summarizes the current state of DRL applications in this field, evaluates their effectiveness, and identifies challenges and future directions.

METHODS

A systematic search was conducted in Google Scholar, PubMed, IEEE Xplore, and Scopus using keywords such as "deep reinforcement learning", "radiation therapy", and "treatment planning". The extracted data were synthesized for an overview and critical analysis.

RESULTS

The application of deep reinforcement learning in radiation therapy plan optimization can generally be divided into three categories: optimizing treatment planning parameters, directly optimizing machine parameters, and adaptive radiotherapy. From the perspective of disease sites, DRL has been applied to cervical cancer, prostate cancer, vestibular schwannoma, and lung cancer. Regarding types of radiation therapy, it has been used in HDRBT, IMRT, SBRT, VMAT, GK, and Cyberknife.

CONCLUSIONS

Deep reinforcement learning technology has played a significant role in advancing the automated optimization of radiation therapy plans. However, there is still a considerable gap before it can be widely applied in clinical settings due to three main reasons: inefficiency, limited methods for quality assessment, and poor interpretability. To address these challenges, significant research opportunities exist in the future, such as constructing evaluators, parallelized training, and exploring continuous action spaces.

Implementation and evaluation of an intelligent automatic treatment planning robot for prostate cancer stereotactic body radiation therapy

PURPOSE

We previously developed a virtual treatment planner (VTP), an artificial intelligence robot, operating a treatment planning system (TPS). Using deep reinforcement learning guided by human knowledge, we trained the VTP to autonomously adjust relevant parameters in treatment plan optimization, similar to a human planner, to generate high-quality plans for prostate cancer stereotactic body radiation therapy (SBRT). This study describes the clinical implementation and evaluation of VTP.

MATERIALS AND METHODS

We integrate VTP with Eclipse TPS using scripting Application Programming Interface. VTP observes dose-volume histograms of relevant structures, decides how to adjust dosimetric constraints, including doses, volumes, and weighting factors, and applies the adjustments to the TPS interface to launch the optimization engine. This process continues until a high-quality plan is achieved. We evaluated VTP's performance using the prostate SBRT case from the 2016 American Association of Medical Dosimetrist/Radiosurgery Society plan study with its plan scoring system, and compared to human-generated plans submitted to the challenge. Using the same scoring system, we also compared the plan quality of 36 prostate SBRT cases (20 planned with IMRT and 16 planned with VMAT) treated at our institution for both VTP and human-generated plans.

RESULTS

In the plan study case, VTP achieved a score of 142.1/150.0, ranking the third in the competition (median 134.6). For the clinical cases, VTP achieved 110.6 ± 6.5 for 20 IMRT plans and 126.2 ± 4.7 for 16 VMAT plans, similar to scores of human-generated plans with 110.4 ± 7.0 for IMRT plans and 125.4 ± 4.4 for VMAT plans. The workflow, plan quality and planning time of VTP were reviewed to be satisfactory by experienced physicists.

CONCLUSION

We successfully implemented VTP to operate a TPS for autonomous human-like treatment planning for prostate SBRT.

Modeling physician's preference in treatment plan approval of stereotactic body radiation therapy of prostate cancer

Objective.Treatment planning of radiation therapy is a time-consuming task. It is desirable to develop automatic planning approaches to generate plans favorable to physicians. The purpose of this study is to develop a deep learning based virtual physician network (VPN) that models physician's preference on plan approval for prostate cancer stereotactic body radiation therapy (SBRT).Approach.VPN takes one planning target volume (PTV) and eight organs at risk structure images, as well as a dose distribution of a plan seeking approval as input. It outputs a probability of approving the plan, and a dose distribution indicating improvements to the input dose. Due to the lack of unapproved plans in our database, VPN is trained using an adversarial framework. 68 prostate cancer patients who received 45Gyin 5-fraction SBRT were selected in this study, with 60 patients for training and cross validation, and 8 patients for independent testing.Main results.The trained VPN was able to differentiate approved and unapproved plans with Area under the curve 0.97 for testing data. For unapproved plans, after applying VPN's suggested dose improvement, the improved dose agreed with ground truth with relative differences2.03±2.17%for PTVD98%,0.49±0.29%for PTVV95%,3.08±2.24%for penile bulbDmean,3.73±2.20%for rectumV50%,and2.06±1.73%for bladderV50%.Significance.VPN was developed to accurately model a physician's preference on plan approval and to provide suggestions on how to improve the dose distribution.

Automatic Inverse Treatment Planning of Gamma Knife Radiosurgery via Deep Reinforcement Learning

PURPOSE

Several inverse planning algorithms have been developed for Gamma Knife (GK) radiosurgery to determine a large number of plan parameters via solving an optimization problem, which typically consists of multiple objectives. The priorities among these objectives need to be repetitively adjusted to achieve a clinically good plan for each patient. This study aimed to achieve automatic and intelligent priority-tuning, by developing a deep reinforcement learning (DRL) based method to model the tuning behaviors of human planners.

METHODS

We built a priority-tuning policy network using deep convolutional neural networks. Its input was a vector composed of multiple plan metrics that were used in our institution for GK plan evaluation. The network can determine which tuning action to take, based on the observed quality of the intermediate plan. We trained the network using an end-to-end DRL framework to approximate the optimal action-value function. A scoring function was designed to measure the plan quality to calculate the received reward of a tuning action.

RESULTS

Vestibular schwannoma was chosen as the test bed in this study. The number of training, validation and testing cases were 5, 5, and 16, respectively. For these three datasets, the average scores of the initial plans obtained with a same initial priority set were 3.63 ± 1.34, 3.83 ± 0.86 and 4.20 ± 0.78, respectively, while can be improved to 5.28 ± 0.23, 4.97 ± 0.44 and 5.22 ± 0.26 through manual priority tuning by human expert planners. Our network achieved competitive results with 5.42 ± 0.11, 5.10 ± 0. 42, 5.28 ± 0.20, respectively.

CONCLUSIONS

Our network can generate GK plans of comparable or slightly higher quality comparing with the plans generated by human planners via manual priority tuning for vestibular schwannoma cases. The network can potentially be incorporated into the clinical workflow as a planning assistance to improve GK planning efficiency and help to reduce plan quality variation caused by inter-planner variability. We also hope that our method can reduce the workload of GK planners and allow them to spend more time on more challenging cases. This article is protected by copyright. All rights reserved.

A hierarchical deep reinforcement learning framework for intelligent automatic treatment planning of prostate cancer intensity modulated radiation therapy

Purpose.We have previously proposed an intelligent automatic treatment planning (IATP) framework that builds a virtual treatment planner network (VTPN) to operate a treatment planning system (TPS) to generate high-quality radiation therapy (RT) treatment plans. While the potential of IATP in automating RT treatment planning has been demonstrated, its poor scalability caused by an almost linear growth of network size with the number of treatment planning parameters (TPPs) is a bottleneck, preventing its application in complicate, but clinically relevant treatment planning problems. The decision-making behavior of the trained network is hard to understand. Motivated by the decision-making process of a human planner, this study proposes a hierarchical IATP framework.Methods and materials.The hierarchical VTPN (HieVTPN) consists of three networks, i.e. Structure-Net, Parameter-Net, and Action-Net. When interacting with a TPS, the networks are employed in a sequential order in each step to decide the structure to adjust, the TPP to adjust for the selected structure, and the specific adjustment manner for the parameter, respectively. We developed an end-to-end hierarchical deep reinforcement learning scheme to simultaneously train the three networks. We then evaluated the effectiveness of the proposed framework in the treatment planning problems for prostate cancer intensity modulated RT (IMRT) and stereotactic body RT (SBRT). We benchmarked the performance of our approach by comparing plans made by VTPN of a parallel architecture, and the human plans submitted for competition in the 2016 American Association of Medical Dosimetrist (AAMD)/Radiosurgery Society (RSS) Plan Study. We analyzed scalability of the network size with respect to the number of TPPs. Numerical experiments were also performed to understand the rationale of the decision-making behaviors of the trained HieVTPN.Results.Both HieVTPNs for prostate IMRT and SBRT were trained successfully using 10 training patient cases and 5 validation cases. For IMRT, HieVTPN was able to generate high-quality plans for 59 testing patient cases that were not included in training process, achieving an average plan score of 8.62 (±0.83), with 9 being the maximal score. The score was comparable to that of the VTPN, 8.45 (±0.48). For SBRT planning, HieVTPN achieved an average plan score of 139.07 on five testing patient cases compared to the score of 132.21 averaged over the human plans summited for competition in AAMD/RSS plan study. Different from VTPN with network size linearly scaling with the number of TPPs, the network size of HieVTPN is almost independent of the number of TPPs. It was also observed that the decision-making behaviors of HieVTPN were understandable and generally agreed with the human experience.Conclusions.With the scalability and explainability, the hierarchical IATP framework is more favorable than the previous framework in terms of handling treatment planning problems involving a large number of TPPs.

Improving efficiency of training a virtual treatment planner network via knowledge-guided deep reinforcement learning for intelligent automatic treatment planning of radiotherapy

PURPOSE

We previously proposed an intelligent automatic treatment planning framework for radiotherapy, in which a virtual treatment planner network (VTPN) is built using deep reinforcement learning (DRL) to operate a treatment planning system (TPS) by adjusting treatment planning parameters in it to generate high-quality plans. We demonstrated the potential feasibility of this idea in prostate cancer intensity-modulated radiation therapy (IMRT). Despite the success, the process to train a VTPN via the standard DRL approach with an ϵ-greedy algorithm was time-consuming. The required training time was expected to grow with the complexity of the treatment planning problem, preventing the development of VTPN for more complicated but clinically relevant scenarios. In this study, we proposed a novel knowledge-guided DRL (KgDRL) approach that incorporated knowledge from human planners to guide the training process to improve the efficiency of training a VTPN.

METHOD

Using prostate cancer IMRT as a test bed, we first summarized a number of rules in the actions of adjusting treatment planning parameters of our in-house TPS. During the training process of VTPN, in addition to randomly navigating the large state-action space, as in the standard DRL approach using the ϵ-greedy algorithm, we also sampled actions defined by the rules. The priority of sampling actions from rules decreased over the training process to encourage VTPN to explore new policy on parameter adjustment that were not covered by the rules. To test this idea, we trained a VTPN using KgDRL and compared its performance with another VTPN trained using the standard DRL approach. Both networks were trained using 10 training patient cases and five additional cases for validation, while another 59 cases were employed for the evaluation purpose.

RESULTS

It was found that both VTPNs trained via KgDRL and standard DRL spontaneously learned how to operate the in-house TPS to generate high-quality plans, achieving plan quality scores of 8.82 (±0.29) and 8.43 (±0.48), respectively. Both VTPNs outperformed treatment planning purely based on the rules, which had a plan score of 7.81 (±1.59). VTPN trained with eight episodes using KgDRL was able to perform similar to that trained using DRL with 100 epochs. The training time was reduced from more than a week to ~13 hrs.

CONCLUSION

The proposed KgDRL framework was effective in accelerating the training process of a VTPN by incorporating human knowledge, which will facilitate the development of VTPN for more complicated treatment planning scenarios.

Prescription Value-Based Automatic Optimization of Importance Factors in Inverse Planning

Objective:

An automatic method for the optimization of importance factors was proposed to improve the efficiency of inverse planning.

Methods:

The automatic method consists of 3 steps: (1) First, the importance factors are automatically and iteratively adjusted based on our proposed penalty strategies. (2) Then, plan evaluation is performed to determine whether the obtained plan is acceptable. (3) If not, a higher penalty is assigned to the unsatisfied objective by multiplying it by a compensation coefficient. The optimization processes are performed alternately until an acceptable plan is obtained or the maximum iteration N_max of step (3) is reached.

Results:

Tested on 2 kinds of clinical cases and compared with manual method, the results showed that the quality of the proposed automatic plan was comparable to, or even better than, the manual plan in terms of the dose–volume histogram and dose distributions.

Conclusions:

The proposed algorithm has potential to significantly improve the efficiency of the existing manual adjustment methods for importance factors and contributes to the development of fully automated planning. Especially, the more the subobjective functions, the more obvious the advantage of our algorithm.

Automation in intensity modulated radiotherapy treatment planning-a review of recent innovations

Radiotherapy treatment planning of complex radiotherapy techniques, such as intensity modulated radiotherapy and volumetric modulated arc therapy, is a resource-intensive process requiring a high level of treatment planner intervention to ensure high plan quality. This can lead to variability in the quality of treatment plans and the efficiency in which plans are produced, depending on the skills and experience of the operator and available planning time. Within the last few years, there has been significant progress in the research and development of intensity modulated radiotherapy treatment planning approaches with automation support, with most commercial manufacturers now offering some form of solution. There is a rapidly growing number of research articles published in the scientific literature on the topic. This paper critically reviews the body of publications up to April 2018. The review describes the different types of automation algorithms, including the advantages and current limitations. Also included is a discussion on the potential issues with routine clinical implementation of such software, and highlights areas for future research.

Many-Objective Particle Swarm Optimization Using Two-Stage Strategy and Parallel Cell Coordinate System

It is a daunting challenge to balance the convergence and diversity of an approximate Pareto front in a many-objective optimization evolutionary algorithm. A novel algorithm, named many-objective particle swarm optimization with the two-stage strategy and parallel cell coordinate system (PCCS), is proposed in this paper to improve the comprehensive performance in terms of the convergence and diversity. In the proposed two-stage strategy, the convergence and diversity are separately emphasized at different stages by a single-objective optimizer and a many-objective optimizer, respectively. A PCCS is exploited to manage the diversity, such as maintaining a diverse archive, identifying the dominance resistant solutions, and selecting the diversified solutions. In addition, a leader group is used for selecting the global best solutions to balance the exploitation and exploration of a population. The experimental results illustrate that the proposed algorithm outperforms six chosen state-of-the-art designs in terms of the inverted generational distance and hypervolume over the DTLZ test suite.

Investigating multi-objective fluence and beam orientation IMRT optimization

Radiation Oncology treatment planning requires compromises to be made between clinical objectives that are invariably in conflict. It would be beneficial to have a 'bird's-eye-view' perspective of the full spectrum of treatment plans that represent the possible trade-offs between delivering the intended dose to the planning target volume (PTV) while optimally sparing the organs-at-risk (OARs). In this work, the authors demonstrate Pareto-aware radiotherapy evolutionary treatment optimization (PARETO), a multi-objective tool featuring such bird's-eye-view functionality, which optimizes fluence patterns and beam angles for intensity-modulated radiation therapy (IMRT) treatment planning. The problem of IMRT treatment plan optimization is managed as a combined monolithic problem, where all beam fluence and angle parameters are treated equally during the optimization. To achieve this, PARETO is built around a powerful multi-objective evolutionary algorithm, called Ferret, which simultaneously optimizes multiple fitness functions that encode the attributes of the desired dose distribution for the PTV and OARs. The graphical interfaces within PARETO provide useful information such as: the convergence behavior during optimization, trade-off plots between the competing objectives, and a graphical representation of the optimal solution database allowing for the rapid exploration of treatment plan quality through the evaluation of dose-volume histograms and isodose distributions. PARETO was evaluated for two relatively complex clinical cases, a paranasal sinus and a pancreas case. The end result of each PARETO run was a database of optimal (non-dominated) treatment plans that demonstrated trade-offs between the OAR and PTV fitness functions, which were all equally good in the Pareto-optimal sense (where no one objective can be improved without worsening at least one other). Ferret was able to produce high quality solutions even though a large number of parameters, such as beam fluence and beam angles, were included in the optimization.

Personalized treatment planning with a model of radiation therapy outcomes for use in multiobjective optimization of IMRT plans for prostate cancer

PURPOSE

To build a new treatment planning approach that extends beyond radiation transport and IMRT optimization by modeling the radiation therapy process and prognostic indicators for more outcome-focused decision making.

METHODS

An in-house treatment planning system was modified to include multiobjective inverse planning, a probabilistic outcome model, and a multi-attribute decision aid. A genetic algorithm generated a set of plans embodying trade-offs between the separate objectives. An influence diagram network modeled the radiation therapy process of prostate cancer using expert opinion, results of clinical trials, and published research. A Markov model calculated a quality adjusted life expectancy (QALE), which was the endpoint for ranking plans.

RESULTS

The Multiobjective Evolutionary Algorithm (MOEA) was designed to produce an approximation of the Pareto Front representing optimal tradeoffs for IMRT plans. Prognostic information from the dosimetrics of the plans, and from patient-specific clinical variables were combined by the influence diagram. QALEs were calculated for each plan for each set of patient characteristics. Sensitivity analyses were conducted to explore changes in outcomes for variations in patient characteristics and dosimetric variables. The model calculated life expectancies that were in agreement with an independent clinical study.

CONCLUSIONS

The radiation therapy model proposed has integrated a number of different physical, biological and clinical models into a more comprehensive model. It illustrates a number of the critical aspects of treatment planning that can be improved and represents a more detailed description of the therapy process. A Markov model was implemented to provide a stronger connection between dosimetric variables and clinical outcomes and could provide a practical, quantitative method for making difficult clinical decisions.

A patient-specific computational model of hypoxia-modulated radiation resistance in glioblastoma using 18F-FMISO-PET

Glioblastoma multiforme (GBM) is a highly invasive primary brain tumour that has poor prognosis despite aggressive treatment. A hallmark of these tumours is diffuse invasion into the surrounding brain, necessitating a multi-modal treatment approach, including surgery, radiation and chemotherapy. We have previously demonstrated the ability of our model to predict radiographic response immediately following radiation therapy in individual GBM patients using a simplified geometry of the brain and theoretical radiation dose. Using only two pre-treatment magnetic resonance imaging scans, we calculate net rates of proliferation and invasion as well as radiation sensitivity for a patient's disease. Here, we present the application of our clinically targeted modelling approach to a single glioblastoma patient as a demonstration of our method. We apply our model in the full three-dimensional architecture of the brain to quantify the effects of regional resistance to radiation owing to hypoxia in vivo determined by [(18)F]-fluoromisonidazole positron emission tomography (FMISO-PET) and the patient-specific three-dimensional radiation treatment plan. Incorporation of hypoxia into our model with FMISO-PET increases the model-data agreement by an order of magnitude. This improvement was robust to our definition of hypoxia or the degree of radiation resistance quantified with the FMISO-PET image and our computational model, respectively. This work demonstrates a useful application of patient-specific modelling in personalized medicine and how mathematical modelling has the potential to unify multi-modality imaging and radiation treatment planning.

A new mathematical approach for handling DVH criteria in IMRT planning

The appropriate handling of planning criteria on the cumulative dose-volume histogram (DVH) is a highly problematic issue in intensity-modulated radiation therapy (IMRT) plan optimization. The nonconvexity of DVH criteria and globality of the resulting optimization problems complicate the design of suitable optimization methods, which feature numerical efficiency, reliable convergence and optimality of the results. This work examines the mathematical structure of DVH criteria and proves the valuable properties of isotonicity/antitonicity, connectedness, invexity and sufficiency of the KKT condition. These properties facilitate the use of efficient and goal-oriented optimization methods. An exemplary algorithmic realization with feasible direction methods gives rise to a functional framework for interactive IMRT planning on DVH criteria. Numerical examples on real world planning cases prove its practical capability.

Clinicians' expectations for gene-driven cancer therapy

A new era of medicine is rapidly approaching, which will change not only pathological diagnosis but also medical decision-making. This paper raises the question of how well prepared doctors are to address the new issues that will soon confront them. The human genome has been completely sequenced and general understanding about cancer biology has increased enormously with understanding that unregulated gene function and complicated changes in signal pathways are related to uncontrolled cell growth. Thus, gene-driven therapy involving alterations to genes are recognized to present new therapy options. This advance will necessitate major changes to the decision-making aspect of physicians. This article focuses on defining the pertinent changes and addressing what they mean for practicing physicians.

Quality assessment for VMAT prostate radiotherapy planning based on data envelopment analysis

The majority of commercial radiotherapy treatment planning systems requires planners to iteratively adjust the plan parameters in order to find a satisfactory plan. This iterative trial-and-error nature of radiotherapy treatment planning results in an inefficient planning process and in order to reduce such inefficiency, plans can be accepted without achieving the best attainable quality. We propose a quality assessment method based on data envelopment analysis (DEA) to address this inefficiency. This method compares a plan of interest to a set of past delivered plans and searches for evidence of potential further improvement. With the assistance of DEA, planners will be able to make informed decisions on whether further planning is required and ensure that a plan is only accepted when the plan quality is close to the best attainable one. We apply the DEA method to 37 prostate plans using two assessment parameters: rectal generalized equivalent uniform dose (gEUD) as the input and D95 (the minimum dose that is received by 95% volume of a structure) of the planning target volume (PTV) as the output. The percentage volume of rectum overlapping PTV is used to account for anatomical variations between patients and is included in the model as a non-discretionary output variable. Five plans that are considered of lesser quality by DEA are re-optimized with the goal to further improve rectal sparing. After re-optimization, all five plans improve in rectal gEUD without clinically considerable deterioration of the PTV D95 value. For the five re-optimized plans, the rectal gEUD is reduced by an average of 1.84 Gray (Gy) with only an average reduction of 0.07 Gy in PTV D95. The results demonstrate that DEA can correctly identify plans with potential improvements in terms of the chosen input and outputs.

Maximizing dosimetric benefits of IMRT in the treatment of localized prostate cancer through multicriteria optimization planning

We examine the quality of plans created using multicriteria optimization (MCO) treatment planning in intensity-modulated radiation therapy (IMRT) in treatment of localized prostate cancer. Nine random cases of patients receiving IMRT to the prostate were selected. Each case was associated with a clinically approved plan created using Corvus. The cases were replanned using MCO-based planning in RayStation. Dose-volume histogram data from both planning systems were presented to 2 radiation oncologists in a blinded evaluation, and were compared at a number of dose-volume points. Both physicians rated all 9 MCO plans as superior to the clinically approved plans (p<10(-5)). Target coverage was equivalent (p = 0.81). Maximum doses to the prostate and bladder and the V50 and V70 to the anterior rectum were reduced in all MCO plans (p<0.05). Treatment planning time with MCO took approximately 60 minutes per case. MCO-based planning for prostate IMRT is efficient and produces high-quality plans with good target homogeneity and sparing of the anterior rectum, bladder, and femoral heads, without sacrificing target coverage.

Adaptive IMRT using a multiobjective evolutionary algorithm integrated with a diffusion-invasion model of glioblastoma

We demonstrate a patient-specific method of adaptive IMRT treatment for glioblastoma using a multiobjective evolutionary algorithm (MOEA). The MOEA generates spatially optimized dose distributions using an iterative dialogue between the MOEA and a mathematical model of tumor cell proliferation, diffusion and response. Dose distributions optimized on a weekly basis using biological metrics have the potential to substantially improve and individualize treatment outcomes. Optimized dose distributions were generated using three different decision criteria for the tumor and compared with plans utilizing standard dose of 1.8 Gy/fraction to the CTV (T2-visible MRI region plus a 2.5 cm margin). The sets of optimal dose distributions generated using the MOEA approach the Pareto Front (the set of IMRT plans that delineate optimal tradeoffs amongst the clinical goals of tumor control and normal tissue sparing). MOEA optimized doses demonstrated superior performance as judged by three biological metrics according to simulated results. The predicted number of reproductively viable cells 12 weeks after treatment was found to be the best target objective for use in the MOEA.

Pareto fronts in clinical practice for pinnacle

PURPOSE

Our aim was to develop a framework to objectively perform treatment planning studies using Pareto fronts. The Pareto front represents all optimal possible tradeoffs among several conflicting criteria and is an ideal tool with which to study the possibilities of a given treatment technique. The framework should require minimal user interaction and should resemble and be applicable to daily clinical practice.

METHODS AND MATERIALS

To generate the Pareto fronts, we used the native scripting language of Pinnacle(3) (Philips Healthcare, Andover, MA). The framework generates thousands of plans automatically from which the Pareto front is generated. As an example, the framework is applied to compare intensity modulated radiation therapy (IMRT) with volumetric modulated arc therapy (VMAT) for prostate cancer patients. For each patient and each technique, 3000 plans are generated, resulting in a total of 60,000 plans. The comparison is based on 5-dimensional Pareto fronts.

RESULTS

Generating 3000 plans for 10 patients in parallel requires on average 96 h for IMRT and 483 hours for VMAT. Using VMAT, compared to IMRT, the maximum dose of the boost PTV was reduced by 0.4 Gy (P=.074), the mean dose in the anal sphincter by 1.6 Gy (P=.055), the conformity index of the 95% isodose (CI(95%)) by 0.02 (P=.005), and the rectal wall V(65 Gy) by 1.1% (P=.008).

CONCLUSIONS

We showed the feasibility of automatically generating Pareto fronts with Pinnacle(3). Pareto fronts provide a valuable tool for performing objective comparative treatment planning studies. We compared VMAT with IMRT in prostate patients and found VMAT had a dosimetric advantage over IMRT.

The use of a multiobjective evolutionary algorithm to increase flexibility in the search for better IMRT plans

PURPOSE

To evaluate how a more flexible and thorough multiobjective search of feasible IMRT plans affects performance in IMRT optimization.

METHODS

A multiobjective evolutionary algorithm (MOEA) was used as a tool to investigate how expanding the search space to include a wider range of penalty functions affects the quality of the set of IMRT plans produced. The MOEA uses a population of IMRT plans to generate new IMRT plans through deterministic minimization of recombined penalty functions that are weighted sums of multiple, tissue-specific objective functions. The quality of the generated plans are judged by an independent set of nonconvex, clinically relevant decision criteria, and all dominated plans are eliminated. As this process repeats itself, better plans are produced so that the population of IMRT plans will approach the Pareto front. Three different approaches were used to explore the effects of expanding the search space. First, the evolutionary algorithm used genetic optimization principles to search by simultaneously optimizing both the weights and tissue-specific dose parameters in penalty functions. Second, penalty function parameters were individually optimized for each voxel in all organs at risk (OARs) in the MOEA. Finally, a heuristic voxel-specific improvement (VSI) algorithm that can be used on any IMRT plan was developed that incrementally improves voxel-specific penalty function parameters for all structures (OARs and targets). Different approaches were compared using the concept of domination comparison applied to the sets of plans obtained by multiobjective optimization.

RESULTS

MOEA optimizations that simultaneously searched both importance weights and dose parameters generated sets of IMRT plans that were superior to sets of plans produced when either type of parameter was fixed for four example prostate plans. The amount of improvement increased with greater overlap between OARs and targets. Allowing the MOEA to search for voxel-specific penalty functions improved results for simple cases with three structures but did not improve results for a more complex case with seven structures. For this modification, the amount of improvement increased with less overlap between OARs and targets. The voxel-specific improvement algorithm improved results for all cases, and its clinical relevance was demonstrated in a complex prostate and a very complex head and neck case.

CONCLUSIONS

Using an evolutionary algorithm as a tool, it was found that allowing more flexibility in the search space enhanced performance. The two strategies of (a) varying the weights and reference doses in the objective function and (b) removing the constraint of equal penalties for all voxels in a structure both generated sets of plans that dominated sets of plans considered to be "Pareto optimal" within the conventional, more limited search space. When considering voxel-specific objectives, the very large search space can lead to convergence problems in the MOEA for complex cases, but this is not an issue for the VSI algorithm.

Introducing multiple treatment plan-based comparison to investigate the performance of gantry angle optimisation (GAO) in IMRT for head and neck cancer

BACKGROUND AND PURPOSE

The purpose of this study was to evaluate the performance of gantry angle optimisation (GAO) compared to equidistant beam geometry for two inverse treatment planning systems (TPSs) by utilising the information obtained from a range of treatment plans.

MATERIAL AND METHODS

The comparison was based on treatment plans generated for four different head and neck (H&N) cancer cases using two inverse treatment planning systems (TPSs); Varian Eclipse™ representing dynamic MLC intensity modulated radiotherapy (IMRT) and Oncentra® Masterplan representing segmented MLC-based IMRT. The patient cases were selected on the criterion of representing different degrees of overlap between the planning target volume (PTV) and the investigated organ at risk, the ipsilateral parotid gland. For each case, a number of 'Pareto optimal' plans were generated in order to investigate the trade-off between the under-dosage to the PTV (V(PTV,D < 95%)) or the decrease in dose homogeneity (D(5)-D(95)) to the PTV as a function of the mean absorbed dose to the ipsilateral parotid gland (<D>(parotid gland)).

RESULTS

For the Eclipse system, GAO had a clear advantage for the cases with smallest overlap (Cases 1 and 2). The set of data points, representing the underlying trade-offs, generated with and without using GAO were, however, not as clearly separated for the cases with larger overlap (Cases 3 and 4). With the OMP system, the difference was less pronounced for all cases. The Eclipse GAO displays the most favourable trade-off for all H&N cases.

CONCLUSIONS

We have found differences in the effectiveness of GAO as compared to equidistant beam geometry, in terms of handling conflicting trade-offs for two commercial inverse TPSs. A comparison, based on a range of treatment plans, as developed in this study, is likely to improve the understanding of conflicting trade-offs and might apply to other thorough comparison techniques.

Quantitative metrics for assessing plan quality

Despite many studies over the last 3 decades that have attempted to explicitly quantify the decision-making process for radiotherapy treatment plan evaluation, judgments of an individual plan's degree of quality are still largely subjective and can show inter- and intra-practitioner variability even if the clinical treatment goals are the same. Several factors conspire to confound the full quantification of treatment plan quality, including uncertainties in dose response of cancerous and normal tissue, the rapid pace of new technology adoption, and the human component of treatment planning. However, new developments in clinical informatics and automation are lowering the bar for developing and implementing quantitative metrics into the treatment planning process. This review discusses general strategies for using quantitative metrics in the treatment planning process and presents a case study in intensity-modulated radiation therapy planning whereby control was established on a variable system via such techniques.

Investigation of effective decision criteria for multiobjective optimization in IMRT

PURPOSE

To investigate how using different sets of decision criteria impacts the quality of intensity modulated radiation therapy (IMRT) plans obtained by multiobjective optimization.

METHODS

A multiobjective optimization evolutionary algorithm (MOEA) was used to produce sets of IMRT plans. The MOEA consisted of two interacting algorithms: (i) a deterministic inverse planning optimization of beamlet intensities that minimizes a weighted sum of quadratic penalty objectives to generate IMRT plans and (ii) an evolutionary algorithm that selects the superior IMRT plans using decision criteria and uses those plans to determine the new weights and penalty objectives of each new plan. Plans resulting from the deterministic algorithm were evaluated by the evolutionary algorithm using a set of decision criteria for both targets and organs at risk (OARs). Decision criteria used included variation in the target dose distribution, mean dose, maximum dose, generalized equivalent uniform dose (gEUD), an equivalent uniform dose (EUD(alpha,beta) formula derived from the linear-quadratic survival model, and points on dose volume histograms (DVHs). In order to quantatively compare results from trials using different decision criteria, a neutral set of comparison metrics was used. For each set of decision criteria investigated, IMRT plans were calculated for four different cases: two simple prostate cases, one complex prostate Case, and one complex head and neck Case.

RESULTS

When smaller numbers of decision criteria, more descriptive decision criteria, or less anti-correlated decision criteria were used to characterize plan quality during multiobjective optimization, dose to OARs and target dose variation were reduced in the final population of plans. Mean OAR dose and gEUD (a = 4) decision criteria were comparable. Using maximum dose decision criteria for OARs near targets resulted in inferior populations that focused solely on low target variance at the expense of high OAR dose. Target dose range, (D(max) - D(min)), decision criteria were found to be most effective for keeping targets uniform. Using target gEUD decision criteria resulted in much lower OAR doses but much higher target dose variation. EUD(alpha,beta) based decision criteria focused on a region of plan space that was a compromise between target and OAR objectives. None of these target decision criteria dominated plans using other criteria, but only focused on approaching a different area of the Pareto front.

CONCLUSIONS

The choice of decision criteria implemented in the MOEA had a significant impact on the region explored and the rate of convergence toward the Pareto front. When more decision criteria, anticorrelated decision criteria, or decision criteria with insufficient information were implemented, inferior populations are resulted. When more informative decision criteria were used, such as gEUD, EUD(alpha,beta), target dose range, and mean dose, MOEA optimizations focused on approaching different regions of the Pareto front, but did not dominate each other. Using simple OAR decision criteria and target EUD(alpha,beta) decision criteria demonstrated the potential to generate IMRT plans that significantly reduce dose to OARs while achieving the same or better tumor control when clinical requirements on target dose variance can be met or relaxed.

When is better best? A multiobjective perspective

PURPOSE

To identify the most informative methods for reporting results of treatment planning comparisons.

METHODS

Seven articles from the past year of International Journal of Radiation Oncology Biology Physics reported on comparisons of treatment plans for IMRT and IMAT. The articles were reviewed to identify methods of comparisons. Decision theoretical concepts were used to evaluate the study methods and highlight those that provide the most information.

RESULTS

None of the studies examined the correlation between objectives. Statistical comparisons provided some information but not enough to provide support for a robust decision analysis.

CONCLUSIONS

The increased use of treatment planning studies to evaluate different methods in radiation therapy requires improved standards for designing the studies and reporting the results.