Fully automated, comprehensive knowledge-based planning for stereotactic radiosurgery: Preclinical validation through blinded physician review

Attention 3D U-NET for dose distribution prediction of high-dose-rate brachytherapy of cervical cancer: Direction modulated brachytherapy tandem applicator

BACKGROUND

Direction Modulated Brachytherapy (DMBT) enables conformal dose distributions. However, clinicians may face challenges in creating viable treatment plans within a fast-paced clinical setting, especially for a novel technology like DMBT, where cumulative clinical experience is limited. Deep learning-based dose prediction methods have emerged as effective tools for enhancing efficiency.

PURPOSE

To develop a voxel-wise dose prediction model using an attention-gating mechanism and a 3D UNET for cervical cancer high-dose-rate (HDR) brachytherapy treatment planning with DMBT six-groove tandems with ovoids or ring applicators.

METHODS

A multi-institutional cohort of 122 retrospective clinical HDR brachytherapy plans treated to a prescription dose in the range of 4.8-7.0 Gy/fraction was used. A DMBT tandem model was constructed and incorporated onto a research version of BrachyVision Treatment Planning System (BV-TPS) as a 3D solid model applicator and retrospectively re-planned all cases by seasoned experts. Those plans were randomly divided into 64:16:20 as training, validating, and testing cohorts, respectively. Data augmentation was applied to the training and validation sets to increase the size by a factor of 4. An attention-gated 3D UNET architecture model was developed to predict full 3D dose distributions based on high-risk clinical target volume (CTVHR) and organs at risk (OARs) contour information. The model was trained using the mean absolute error loss function, Adam optimization algorithm, a learning rate of 0.001, 250 epochs, and a batch size of eight. In addition, a baseline UNET model was trained similarly for comparison. The model performance was evaluated on the testing dataset by analyzing the outcomes in terms of mean dose values and derived dose-volume-histogram indices from 3D dose distributions and comparing the generated dose distributions against the ground-truth dose distributions using dose statistics and clinically meaningful dosimetric indices.

RESULTS

The proposed attention-gated 3D UNET model showed competitive accuracy in predicting 3D dose distributions that closely resemble the ground-truth dose distributions. The average values of the mean absolute errors were 1.82 ± 29.09 Gy (vs. 6.41 ± 20.16 Gy for a baseline UNET) in CTVHR, 0.89 ± 1.25 Gy (vs. 0.94 ± 3.96 Gy for a baseline UNET) in the bladder, 0.33 ± 0.67 Gy (vs. 0.53 ± 1.66 Gy for a baseline UNET) in the rectum, and 0.55 ± 1.57 Gy (vs. 0.76 ± 2.89 Gy for a baseline UNET) in the sigmoid. The results showed that the mean absolute error (MAE) for the bladder, rectum, and sigmoid were 0.22 ± 1.22 Gy (3.62%) (p = 0.015), 0.21 ± 1.06 Gy (2.20%) (p = 0.172), and -0.03 ± 0.54 Gy (1.13%) (p = 0.774), respectively. The MAE for D90, V100%, and V150% of the CTVHR were 0.46 ± 2.44 Gy (8.14%) (p = 0.018), 0.57 ± 11.25% (5.23%) (p = 0.283), and -0.43 ± 19.36% (4.62%) (p = 0.190), respectively. The proposed model needs less than 5 s to predict a full 3D dose distribution of 64 × 64 × 64 voxels for any new patient plan, thus making it sufficient for near real-time applications and aiding with decision-making in the clinic.

CONCLUSIONS

Attention gated 3D-UNET model demonstrated a capability in predicting voxel-wise dose prediction, in comparison to 3D UNET, for DMBT intracavitary brachytherapy planning. The proposed model could be used to obtain dose distributions for near real-time decision-making before DMBT planning and quality assurance. This will guide future automated planning, making the workflow more efficient and clinically viable.

Effects of model size and composition on quality of head-and-neck knowledge-based plans

PURPOSE

Knowledge-based planning (KBP) aims to automate and standardize treatment planning. New KBP users are faced with many questions: How much does model size matter, and are multiple models needed to accommodate specific physician preferences? In this study, six head-and-neck KBP models were trained to address these questions.

METHODS

The six models differed in training size and plan composition: The KBPFull (n = 203 plans), KBP101 (n = 101), KBP50 (n = 50), and KBP25 (n = 25) were trained with plans from two head-and-neck physicians. KBPA and KBPB each contained n = 101 plans from only one physician, respectively. An independent set of 39 patients treated to 6000-7000 cGy by a third physician was re-planned with all KBP models for validation. Standard head-and-neck dosimetric parameters were used to compare resulting plans. KBPFull plans were compared to the clinical plans to evaluate overall model quality. Additionally, clinical and KBPFull plans were presented to another physician for blind review. Dosimetric comparison of KBPFull against KBP101 , KBP50 , and KBP25 investigated the effect of model size. Finally, KBPA versus KBPB tested whether training KBP models on plans from one physician only influences the resulting output. Dosimetric differences were tested for significance using a paired t-test (p < 0.05).

RESULTS

Compared to manual plans, KBPFull significantly increased PTV Low D95% and left parotid mean dose but decreased dose cochlea, constrictors, and larynx. The physician preferred the KBPFull plan over the manual plan in 20/39 cases. Dosimetric differences between KBPFull , KBP101 , KBP50 , and KBP25 plans did not exceed 187 cGy on aggregate, except for the cochlea. Further, average differences between KBPA and KBPB were below 110 cGy.

CONCLUSIONS

Overall, all models were shown to produce high-quality plans. Differences between model outputs were small compared to the prescription. This indicates only small improvements when increasing model size and minimal influence of the physician when choosing treatment plans for training head-and-neck KBP models.

A comparative study of deep learning-based knowledge-based planning methods for 3D dose distribution prediction of head and neck

PURPOSE

In this paper, we compare four novel knowledge-based planning (KBP) algorithms using deep learning to predict three-dimensional (3D) dose distributions of head and neck plans using the same patients' dataset and quantitative assessment metrics.

METHODS

A dataset of 340 oropharyngeal cancer patients treated with intensity-modulated radiation therapy was used in this study, which represents the AAPM OpenKBP - 2020 Grand Challenge dataset. Four 3D convolutional neural network architectures were built. The models were trained on 64% of the data set and validated on 16% for voxel-wise dose predictions: U-Net, attention U-Net, residual U-Net (Res U-Net), and attention Res U-Net. The trained models were then evaluated for their performance on a test data set (20% of the data) by comparing the predicted dose distributions against the ground-truth using dose statistics and dose-volume indices.

RESULTS

The four KBP dose prediction models exhibited promising performance with an averaged mean absolute dose error within the body contour <3 Gy on 68 plans in the test set. The average difference in predicting the D99 index for all targets was 0.92 Gy (p = 0.51) for attention Res U-Net, 0.94 Gy (p = 0.40) for Res U-Net, 2.94 Gy (p = 0.09) for attention U-Net, and 3.51 Gy (p = 0.08) for U-Net. For the OARs, the values for theD m a x ${D_{max}}$ andD m e a n ${D_{mean}}$ indices were 2.72 Gy (p < 0.01) for attention Res U-Net, 2.94 Gy (p < 0.01) for Res U-Net, 1.10 Gy (p < 0.01) for attention U-Net, 0.84 Gy (p < 0.29) for U-Net.

CONCLUSION

All models demonstrated almost comparable performance for voxel-wise dose prediction. KBP models that employ 3D U-Net architecture as a base could be deployed for clinical use to improve cancer patient treatment by creating plans with consistent quality and making the radiotherapy workflow more efficient.

Can the use of knowledge-based planning systems improve stereotactic radiotherapy planning? A systematic review

Introduction:

This study aimed to systematically review the literature to synthesise and summarise whether using knowledge-based planning (KBP) can improve the planning of stereotactic radiotherapy treatments.

Methods:

A systematic literature search was carried out using Medline, Scopus and Cochrane databases to evaluate the use of KBP planning in stereotactic radiotherapy. Three hundred twenty-five potential studies were identified and screened to find 25 relevant studies.

Results:

Twenty-five studies met the inclusion criteria. Where a commercial KBP was used, 72.7% of studies reported a quality improvement, and 45.5% reported a reduction in planning time. There is evidence that when used as a quality control tool, KBP can highlight stereotactic plans that need revision. In studies that use KBP as the starting point for radiotherapy planning optimisation, the radiotherapy plans generated are typically equal to or superior to those planned manually.

Conclusions:

There is evidence that KBP has the potential to improve the quality and speed of stereotactic radiotherapy planning. Further research is required to accurately quantify such systems’ quality improvements and time savings. Notably, there has been little research into their use for prostate, spinal or liver stereotactic radiotherapy, and research in these areas would be desirable. It is recommended that future studies use the ICRU 91 level 2 reporting format and that blinded physician review could add a qualitative assessment of KBP system performance.

Input feature design and its impact on the performance of deep learning models for predicting fluence maps in intensity-modulated radiation therapy

Objective. Deep learning (DL) models for fluence map prediction (FMP) have great potential to reduce treatment planning time in intensity-modulated radiation therapy (IMRT) by avoiding the lengthy inverse optimization process. This study aims to improve the rigor of input feature design in a DL-FMP model by examining how different designs of input features influence model prediction performance.Approach. This study included 231 head-and-neck intensity-modulated radiation therapy patients. Three input feature designs were investigated. The first design (D1) assumed that information of all critical structures from all beam angles should be combined to predict fluence maps. The second design (D2) assumed that local anatomical information was sufficient for predicting radiation intensity of a beamlet at a respective beam angle. The third design (D3) assumed the need for both local anatomical information and inter-beam modulation to predict radiation intensity values of the beamlets that intersect at a voxel. For each input design, we tailored the DL model accordingly. All models were trained using the same set of ground truth plans (GT plans). The plans generated by DL models (DL plans) were analyzed using key dose-volume metrics. One-way ANOVA with multiple comparisons correction (Bonferroni method) was performed (significance level = 0.05).Main results. For PTV-related metrics, all DL plans had significantly higher maximum dose (p < 0.001), conformity index (p < 0.001), and heterogeneity index (p < 0.001) compared to GT plans, with D2 being the worst performer. Meanwhile, except for cord+5 mm (p < 0.001), DL plans of all designs resulted in OAR dose metrics that are comparable to those of GT plans.Significance. Local anatomical information contains most of the information that DL models need to predict fluence maps for clinically acceptable OAR sparing. Input features from beam angles are needed to achieve the best PTV coverage. These results provide valuable insights for further improvement of DL-FMP models and DL models in general.

Attention-aware 3D U-Net convolutional neural network for knowledge-based planning 3D dose distribution prediction of head-and-neck cancer

PURPOSE

Deep learning-based knowledge-based planning (KBP) methods have been introduced for radiotherapy dose distribution prediction to reduce the planning time and maintain consistent high-quality plans. This paper presents a novel KBP model using an attention-gating mechanism and a three-dimensional (3D) U-Net for intensity-modulated radiation therapy (IMRT) 3D dose distribution prediction in head-and-neck cancer.

METHODS

A total of 340 head-and-neck cancer plans, representing the OpenKBP-2020 AAPM Grand Challenge data set, were used in this study. All patients were treated with the IMRT technique and a dose prescription of 70 Gy. The data set was randomly divided into 64%/16%/20% as training/validation/testing cohorts. An attention-gated 3D U-Net architecture model was developed to predict full 3D dose distribution. The developed model was trained using the mean-squared error loss function, Adam optimization algorithm, a learning rate of 0.001, 120 epochs, and batch size of 4. In addition, a baseline U-Net model was also similarly trained for comparison. The model performance was evaluated on the testing data set by comparing the generated dose distributions against the ground-truth dose distributions using dose statistics and clinical dosimetric indices. Its performance was also compared to the baseline model and the reported results of other deep learning-based dose prediction models.

RESULTS

The proposed attention-gated 3D U-Net model showed high capability in accurately predicting 3D dose distributions that closely replicated the ground-truth dose distributions of 68 plans in the test set. The average value of the mean absolute dose error was 2.972 ± 1.220 Gy (vs. 2.920 ± 1.476 Gy for a baseline U-Net) in the brainstem, 4.243 ± 1.791 Gy (vs. 4.530 ± 2.295 Gy for a baseline U-Net) in the left parotid, 4.622 ± 1.975 Gy (vs. 4.223 ± 1.816 Gy for a baseline U-Net) in the right parotid, 3.346 ± 1.198 Gy (vs. 2.958 ± 0.888 Gy for a baseline U-Net) in the spinal cord, 6.582 ± 3.748 Gy (vs. 5.114 ± 2.098 Gy for a baseline U-Net) in the esophagus, 4.756 ± 1.560 Gy (vs. 4.992 ± 2.030 Gy for a baseline U-Net) in the mandible, 4.501 ± 1.784 Gy (vs. 4.925 ± 2.347 Gy for a baseline U-Net) in the larynx, 2.494 ± 0.953 Gy (vs. 2.648 ± 1.247 Gy for a baseline U-Net) in the PTV_70, and 2.432 ± 2.272 Gy (vs. 2.811 ± 2.896 Gy for a baseline U-Net) in the body contour. The average difference in predicting the D₉₉ value for the targets (PTV_70, PTV_63, and PTV_56) was 2.50 ± 1.77 Gy. For the organs at risk, the average difference in predicting the D m a x ${D_{max}}$ (brainstem, spinal cord, and mandible) and D m e a n ${D_{mean}}$ (left parotid, right parotid, esophagus, and larynx) values was 1.43 ± 1.01 and 2.44 ± 1.73 Gy, respectively. The average value of the homogeneity index was 7.99 ± 1.45 for the predicted plans versus 5.74 ± 2.95 for the ground-truth plans, whereas the average value of the conformity index was 0.63 ± 0.17 for the predicted plans versus 0.89 ± 0.19 for the ground-truth plans. The proposed model needs less than 5 s to predict a full 3D dose distribution of 64 × 64 × 64 voxels for a new patient that is sufficient for real-time applications.

CONCLUSIONS

The attention-gated 3D U-Net model demonstrated a capability in predicting accurate 3D dose distributions for head-and-neck IMRT plans with consistent quality. The prediction performance of the proposed model was overall superior to a baseline standard U-Net model, and it was also competitive to the performance of the best state-of-the-art dose prediction method reported in the literature. The proposed model could be used to obtain dose distributions for decision-making before planning, quality assurance of planning, and guiding-automated planning for improved plan consistency, quality, and planning efficiency.

A hyperparameter-tuning approach to automated inverse planning

BACKGROUND

In current practice, radiotherapy inverse planning often requires treatment planners to modify multiple parameters in the treatment planning system's objective function to produce clinically acceptable plans. Due to the manual steps in this process, plan quality can vary depending on the planning time available and the planner's skills.

PURPOSE

This study investigates the feasibility of two hyperparameter-tuning methods for automated inverse planning. Because this framework does not train a model on previously-optimized plans, it can be readily adapted to practice pattern changes, and the resulting plan quality is not limited by that of a training cohort.

METHODS

We retrospectively selected 10 patients who received lung stereotactic body radiation therapy using manually-generated clinical plans. We implemented random sampling and Bayesian optimization to automatically tune objective function parameters using linear-quadratic utility functions based on 11 clinical goals. Normalizing all plans to deliver a minimum dose of 48 Gy to 95% of the planning target volume, we compared plan quality for the automatically-generated plans to the manually-generated plans. We also investigated the impact of iteration count on the automatically-generated plans, comparing planning time and plan utility for randomized and Bayesian plans with and without stopping criteria.

RESULTS

Without stopping criteria, the median planning time was 1.9 and 2.3 hours for randomized and Bayesian plans, respectively. The organ-at-risk doses in the randomized and Bayesian plans had a median percent difference (MPD) of 48.7% and 60.4% below clinical dose limits and an MPD of 2.8% and 3.3% below clinical plan doses. With stopping criteria, the utility decreased by an MPD of 5.3% and 3.9% for randomized and Bayesian plans, but the median planning time was reduced to 0.5 and 0.7 hours, and the organ-at-risk doses still had an MPD of 42.9% and 49.7% below clinical dose limits and an MPD of 0.3% and 1.8% below clinical plan doses.

CONCLUSIONS

This study demonstrates that hyperparameter-tuning approaches to automated inverse planning can reduce the treatment planner's active planning time with plan quality that is similar to or better than manually-generated plans. This article is protected by copyright. All rights reserved.

Artificial intelligence applications in intensity modulated radiation treatment planning: an overview

Artificial intelligence (AI) refers to methods that improve and automate challenging human tasks by systematically capturing and applying relevant knowledge in these tasks. Over the past decades, a number of approaches have been developed to address different types and needs of system intelligence ranging from search strategies to knowledge representation and inference to robotic planning. In the context of radiation treatment planning, multiple AI approaches may be adopted to improve the planning quality and efficiency. For example, knowledge representation and inference methods may improve dose prescription by integrating and reasoning about the domain knowledge described in many clinical guidelines and clinical trials reports. In this review, we will focus on the most studied AI approach in intensity modulated radiation therapy (IMRT)/volumetric modulated arc therapy (VMAT)-machine learning (ML) and describe our recent efforts in applying ML to improve the quality, consistency, and efficiency of IMRT/VMAT planning. With the available high-quality data, we can build models to accurately predict critical variables for each step of the planning process and thus automate and improve its outcomes. Specific to the IMRT/VMAT planning process, we can build models for each of the four critical components in the process: dose-volume histogram (DVH), Dose, Fluence, and Human Planner. These models can be divided into two general groups. The first group focuses on encoding prior experience and knowledge through ML and more recently deep learning (DL) from prior clinical plans and using these models to predict the optimal DVH (DVH prediction model), or 3D dose distribution (dose prediction model), or fluence map (fluence map model). The goal of these models is to reduce or remove the trial-and-error process and guarantee consistently high-quality plans. The second group of models focuses on mimicking human planners' decision-making process (planning strategy model) during the iterative adjustments/guidance of the optimization engine. Each critical step of the IMRT/VMAT treatment planning process can be improved and automated by AI methods. As more training data becomes available and more sophisticated models are developed, we can expect that the AI methods in treatment planning will continue to improve accuracy, efficiency, and robustness.

Wide-Scale Clinical Implementation of Knowledge-Based Planning: An Investigation of Workforce Efficiency, Need for Post-automation Refinement, and Data-Driven Model Maintenance

PURPOSE

Our purpose was to investigate the effect of automated knowledge-based planning (KBP) on real-world clinical workflow efficiency, assess whether manual refinement of KBP plans improves plan quality across multiple disease sites, and develop a data-driven method to periodically improve KBP automated planning routines.

METHODS AND MATERIALS

Using clinical knowledge-based automated planning routines for prostate, prostatic fossa, head and neck, and hypofractionated lung disease sites in a commercial KBP solution, workflow efficiency was compared in terms of planning time in a pre-KBP (n = 145 plans) and post-KBP (n = 503) patient cohort. Post-KBP, planning was initialized with KBP (KBP-only) and subsequently manually refined (KBP + human). Differences in planning time were tested for significance using a 2-tailed Mann-Whitney U test (P < .05, null hypothesis: planning time unchanged). Post-refinement plan quality was assessed using site-specific dosimetric parameters of the original KBP-only plan versus KBP + human; 2-tailed paired t test quantified statistical significance (Bonferroni-corrected P < .05, null hypothesis: no dosimetric difference after refinement). If KBP + human significantly improved plans across the cohort, optimization objectives were changed to create an updated KBP routine (KBP'). Patients were replanned with KBP' and plan quality was compared with KBP + human as described previously.

RESULTS

KBP significantly reduced planning time in all disease sites: prostate (median: 7.6 hrs → 2.1 hrs; P < .001), prostatic fossa (11.1 hrs → 3.7 hrs; P = .001), lung (9.9 hrs → 2.0 hrs; P < .001), and head and neck (12.9 hrs → 3.5 hrs; P <.001). In prostate, prostatic fossa, and lung disease sites, organ-at-risk dose changes in KBP + human versus KBP-only were minimal (<1% prescription dose). In head and neck, KBP + human did achieve clinically relevant dose reductions in some parameters. The head and neck routine was updated (KBP'_HN) to incorporate dose improvements from manual refinement. The only significant dosimetric differences to KBP + human after replanning with KBP'_HN were in favor of the new routine.

CONCLUSIONS

KBP increased clinical efficiency by significantly reducing planning time. On average, human refinement offered minimal dose improvements over KBP-only plans. In the single disease site where KBP + human was superior to KBP-only, differences were eliminated by adjusting optimization parameters in a revised KBP routine.

Clinical Acceptability of Automated Radiation Treatment Planning for Head and Neck Cancer Using the Radiation Planning Assistant

PURPOSE

Radiation treatment planning for head and neck cancer is a complex process with much variability; automated treatment planning is a promising option to improve plan quality and efficiency. This study compared radiation plans generated from a fully automated radiation treatment planning system to plans generated manually that had been clinically approved and delivered.

METHODS AND MATERIALS

The study cohort consisted of 50 patients treated by a specialized head and neck cancer team at a tertiary care center. An automated radiation treatment planning system, the Radiation Planning Assistant, was used to create autoplans for all patients using their original, approved contours. Common dose-volume histogram (DVH) criteria were used to compare the quality of autoplans to the clinical plans. Fourteen radiation oncologists, each from a different institution, then reviewed and compared the autoplans and clinical plans in a blinded fashion.

RESULTS

Autoplans and clinical plans were very similar with regard to DVH metrics for coverage and critical structure constraints. Physician reviewers found both the clinical plans and autoplans acceptable for use; overall, 78% of the clinical plans and 88% of the autoplans were found to be usable as is (without any edits). When asked to choose which plan would be preferred for approval, 27% of physician reviewers selected the clinical plan, 47% selected the autoplan, 25% said both were equivalent, and 0% said neither. Hence, overall, 72% of physician reviewers believed the autoplan or either the clinical or autoplan was preferable.

CONCLUSIONS

Automated radiation treatment planning creates consistent, clinically acceptable treatment plans that meet DVH criteria and are found to be appropriate on physician review.

Artificial intelligence in radiotherapy: a technological review

Radiation therapy (RT) is widely used to treat cancer. Technological advances in RT have occurred in the past 30 years. These advances, such as three-dimensional image guidance, intensity modulation, and robotics, created challenges and opportunities for the next breakthrough, in which artificial intelligence (AI) will possibly play important roles. AI will replace certain repetitive and labor-intensive tasks and improve the accuracy and consistency of others, particularly those with increased complexity because of technological advances. The improvement in efficiency and consistency is important to manage the increasing cancer patient burden to the society. Furthermore, AI may provide new functionalities that facilitate satisfactory RT. The functionalities include superior images for real-time intervention and adaptive and personalized RT. AI may effectively synthesize and analyze big data for such purposes. This review describes the RT workflow and identifies areas, including imaging, treatment planning, quality assurance, and outcome prediction, that benefit from AI. This review primarily focuses on deep-learning techniques, although conventional machine-learning techniques are also mentioned.

Development of integrated prompt gamma imaging and positron emission tomography system for in vivo 3-D dose verification: a Monte Carlo study

An accurate knowledge of in vivo proton dose distribution is key to fully utilizing the potential advantages of proton therapy. Two representative indirect methods for in vivo range verification, namely, prompt gamma (PG) imaging and positron emission tomography (PET), are available. This study proposes a PG-PET system that combines the advantages of these two methods and presents detector geometry and background reduction techniques optimized for the PG-PET system. The characteristics of the secondary radiations emitted by a water phantom by interaction with a 150 MeV proton beam were analysed using Geant4.10.00, and the 2-D PG distributions were obtained and assessed for different detector geometries. In addition, the energy window (EW), depth-of-interaction (DOI), and time-of-flight (TOF) techniques are proposed as the background reduction techniques. To evaluate the performance of the PG-PET system, the 3-D dose distribution in the water phantom caused by two proton beams of energies 80 MeV and 100 MeV was verified using 16 optimal detectors. The thickness of the parallel-hole tungsten collimator of pitch 8 mm and width 7 mm was determined as 200 mm, and that of the GAGG scintillator was determined as 30 mm, by an optimization study. Further, 3-7 MeV and 2-7 MeV were obtained as the optimal EWs when the DOI and both the DOI and TOF techniques were applied for data processing, respectively; the detector performances were improved by about 38% and 167%, respectively, compared with that when applying only the 3-5 MeV EW. In this study, we confirmed that the PG distribution can be obtained by simply combining the 2-D parallel hole collimator and the PET detector module. In the future, we will develop an accurate 3-D dose evaluation technique using deep learning algorithms based on the image sets of dose, PG, and PET distributions for various proton energies.

[Stereotactic body radiotherapy. How to better protect normal tissues?]

The use of stereotactic body radiotherapy (SBRT) has increased rapidly over the past decade. Optimal preservation of normal tissues is a major issue because of their high sensitivity to high doses per session. Extreme hypofractionation can convert random errors into systematic errors. Optimal preservation of organs at risk requires first of all a rigorous implementation of this technique according to published guidelines. The robustness of the imaging modalities used for planning, and training medical and paramedical staff are an integral part of these guidelines too. The choice of SBRT indications, dose fractionation, dose heterogeneity, ballistics, are also means of optimizing the protection of normal tissues. Non-coplanarity and tracking of moving targets allow dosimetric improvement in some clinical settings. Automatic planning could also improve normal tissue protection. Adaptive SBRT, with new image guided radiotherapy modalities such as MRI, could further reduce the risk of toxicity.

Automated 4π radiotherapy treatment planning with evolving knowledge-base

PURPOSE

Non-coplanar 4π radiotherapy generalizes intensity modulated radiation therapy (IMRT) to automate beam geometry selection but requires complicated hyperparameter tuning to attain superior plan quality, which can be tedious and inconsistent. In this study, a fully automated 4π treatment planning was developed using evolving knowledge-base (EKB) planning guided by dose prediction.

METHODS

Twenty 4π lung and twenty 4π head and neck (HN) cases were included. A statistical voxel dose learning model was initially trained on low-quality plans created using generic hyperparameter templates without manual tuning. To improve the automated plan quality without being limited by the training data quality, a new 4π optimization problem was formulated to include a one-sided penalty on the organ-at-risk (OAR) dose deviation from the predicted dose. This directional OAR penalty encourages superior OAR sparing. The fast iterative shrinkage-thresholding algorithm (FISTA) was used to solve the large-scale beam orientation optimization problem. With the improved plans, new predictions were created to guide the next loop of EKB planning for a total of 10 loops. Plan quality was evaluated using a plan quality metric (PQM) points system based on clinical dose constraints and compared with automated planning approaches guided by manual high-quality plans using all non-coplanar beams, automated plans using individually evolved targeted dose, and manually created 4π plans.

RESULTS

For the lung cases, the final EKB plans had significantly higher PQM than manually created 4π (+2.60%). The improvements plateaued after the third loop. The final HN EKB plans and manually created 4π plans had comparable PQMs, but had lower PQM compared to automated plans using a high-quality training set (-3.00% and -4.44%, respectively). The PQM consistently increased up to the sixth loop. Individually evolved plans were able to improve the plan quality from initial condition due to the one-sided cost function but the 60% of them were trapped in undesired local minima that were substantially worse than their corresponding EKB plans.

CONCLUSION

Evolving knowledge-base planning is a novel automated planning technique guided by the predicted three-dimensional dose distribution, which can evolve from low-quality plans. EKB allows new beams to be used in the automated planning workflow for superior plan quality.

Applications and limitations of machine learning in radiation oncology

Machine learning approaches to problem-solving are growing rapidly within healthcare, and radiation oncology is no exception. With the burgeoning interest in machine learning comes the significant risk of misaligned expectations as to what it can and cannot accomplish. This paper evaluates the role of machine learning and the problems it solves within the context of current clinical challenges in radiation oncology. The role of learning algorithms within the workflow for external beam radiation therapy are surveyed, considering simulation imaging, multimodal fusion, image segmentation, treatment planning, quality assurance, and treatment delivery and adaptation. For each aspect, the clinical challenges faced, the learning algorithms proposed, and the successes and limitations of various approaches are analyzed. It is observed that machine learning has largely thrived on reproducibly mimicking conventional human-driven solutions with more efficiency and consistency. On the other hand, since algorithms are generally trained using expert opinion as ground truth, machine learning is of limited utility where problems or ground truths are not well-defined, or if suitable measures of correctness are not available. As a result, machines may excel at replicating, automating and standardizing human behaviour on manual chores, meanwhile the conceptual clinical challenges relating to definition, evaluation, and judgement remain in the realm of human intelligence and insight.

Performance Comparison of Knowledge-Based Dose Prediction Techniques Based on Limited Patient Data

Purpose:

The accuracy of dose prediction is essential for knowledge-based planning and automated planning techniques. We compare the dose prediction accuracy of 3 prediction methods including statistical voxel dose learning, spectral regression, and support vector regression based on limited patient training data.

Methods:

Statistical voxel dose learning, spectral regression, and support vector regression were used to predict the dose of noncoplanar intensity-modulated radiation therapy (4π) and volumetric-modulated arc therapy head and neck, 4π lung, and volumetric-modulated arc therapy prostate plans. Twenty cases of each site were used for k-fold cross-validation, with k = 4. Statistical voxel dose learning bins voxels according to their Euclidean distance to the planning target volume and uses the median to predict the dose of new voxels. Distance to the planning target volume, polynomial combinations of the distance components, planning target volume, and organ at risk volume were used as features for spectral regression and support vector regression. A total of 28 features were included. Principal component analysis was performed on the input features to test the effect of dimension reduction. For the coplanar volumetric-modulated arc therapy plans, separate models were trained for voxels within the same axial slice as planning target volume voxels and voxels outside the primary beam. The effect of training separate models for each organ at risk compared to all voxels collectively was also tested. The mean squared error was calculated to evaluate the voxel dose prediction accuracy.

Results:

Statistical voxel dose learning using separate models for each organ at risk had the lowest root mean squared error for all sites and modalities: 3.91 Gy (head and neck 4π), 3.21 Gy (head and neck volumetric-modulated arc therapy), 2.49 Gy (lung 4π), and 2.35 Gy (prostate volumetric-modulated arc therapy). Compared to using the original features, principal component analysis reduced the 4π prediction error for head and neck spectral regression (−43.9%) and support vector regression (−42.8%) and lung support vector regression (−24.4%) predictions. Principal component analysis was more effective in using all/most of the possible principal components. Separate organ at risk models were more accurate than training on all organ at risk voxels in all cases.

Conclusion:

Compared with more sophisticated parametric machine learning methods with dimension reduction, statistical voxel dose learning is more robust to patient variability and provides the most accurate dose prediction method.

Automated Closed- and Open-Loop Validation of Knowledge-Based Planning Routines Across Multiple Disease Sites

PURPOSE

Knowledge-based planning (KBP) clinical implementation necessitates significant upfront effort, even within a single disease site. The purpose of this study was to demonstrate an efficient method for clinicians to assess the noninferiority of KBP across multiple disease sites and estimate any systematic dosimetric differences after implementation. We sought to establish these endpoints in a plurality of previously treated patients (validation set) with both closed-loop (training set overlapping validation set) and open-loop (independent training set) KBP routines.

METHODS AND MATERIALS

We identified 53 prostate, 24 prostatic fossa, 54 hypofractionated lung, and 52 head and neck patients treated with volumetric modulated arc therapy in the year directly preceding our clinic's broad adoption of RapidPlan (Varian Medical Systems, Palo Alto, CA). Using the Varian Eclipse Scripting API, our program takes as input a list of patients, then performs semiautomated structure matching, fully automated RapidPlan-driven optimization, and plan comparison. All plans were normalized to the planning target volume (PTV) D_95% = 100%. Dose metric differences (ΔD_x = D_x,clinical - D_x,KBP) were computed for standard PTV and organ-at-risk (OAR) dose-volume histogram parameters across disease sites. A 2-tailed paired t test quantified statistical significance (P < .001).

RESULTS

Statistically significant organ dose-volume histogram improvements were observed in the KBP cohort: the rectum, bladder, and penile bulb in prostate/prostatic fossa; and the larynx, esophagus, cricopharyngeus, parotid glands, and cochlea in head and neck. No OAR dose metric was statistically worse in any KBP sample. PTV ΔD_1% increases in prostatic fossa were deemed acceptable given organ-sparing gains. PTV ΔD_1% and internal target volume ΔD_99% increase for the lung was by design owing to the prescription normalization variance in the pre-KBP lung sample.

CONCLUSIONS

Our automated method showed multiple disease sites' KBP routines to be noninferior to manual planning, with statistically significant superiority in some aspects of OAR sparing. This method is applicable to any institution implementing either closed-loop or open-loop KBP autoplanning routines.

Incorporating Case-Based Reasoning for Radiation Therapy Knowledge Modeling: A Pelvic Case Study

Knowledge models in radiotherapy capture the relation between patient anatomy and dosimetry to provide treatment planning guidance. When treatment schemes evolve, existing models struggle to predict accurately. We propose a case-based reasoning framework designed to handle novel anatomies that are of same type but vary beyond original training samples. A total of 105 pelvic intensity-modulated radiotherapy cases were analyzed. Eighty cases were prostate cases while the other 25 were prostate-plus-lymph-node cases. We simulated 4 scenarios: Scarce scenario, Semiscarce scenario, Semiample scenario, and Ample scenario. For the Scarce scenario, a multiple stepwise regression model was trained using 85 cases (80 prostate, 5 prostate-plus-lymph-node). The proposed workflow started with evaluating the feature novelty of new cases against 5 training prostate-plus-lymph-node cases using leverage statistic. The case database was composed of a 5-case dose atlas. Case-based dose prediction was compared against the regression model prediction using sum of squared residual. Mean sum of squared residual of case-based and regression predictions for the bladder of 13 identified outliers were 0.174 ± 0.166 and 0.459 ± 0.508, respectively (P = .0326). For the rectum, the respective mean sum of squared residuals were 0.103 ± 0.120 and 0.150 ± 0.171 for case-based and regression prediction (P = .1972). By retaining novel cases, under the Ample scenario, significant statistical improvement was observed over the Scarce scenario (P = .0398) for the bladder model. We expect that the incorporation of case-based reasoning that judiciously applies appropriate predictive models could improve overall prediction accuracy and robustness in clinical practice.

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.