Highly Efficient Training, Refinement, and Validation of a Knowledge-based Planning Quality-Control System for Radiation Therapy Clinical Trials

Can knowledge-based planning models validated on ethnically diverse patients lead to global standardisation of external beam radiation therapy for locally advanced cervix cancer?

BACKGROUND AND PURPOSE

Knowledge-based planning (KBP) can consistently and efficiently create high-quality Volumetric Arc Therapy (VMAT) plans for cervix cancer. This study describes the cross-validation of two KBP models on geographically distinct populations and their comparison to manual plans from 67 centers. The purpose was to determine the universal applicability of a generic KBP model.

MATERIALS AND METHODS

Based on the EMBRACE-II protocol, two KBP models were developed at Tata Memorial Centre, India and Aarhus University Hospital, Denmark using respective patient plans. The KBP models were exchanged between three institutions with different geo-ethnic populations and validated on reference manual plans of 20 node-positive and 20 node-negative patients. Additionally, one patient case was manually planned by 67 centres. These manual treatment plans were compared to the two KBP model plans using a score out of 80, based on 20 DVH parameters.

RESULTS

The manual and the KBP plans adhered to the EMBRACE II protocol. OAR sparing in KBP plans was similar or slightly improved as compared to the manual plans. The differences between the medians of manual and either KBP model plans were significant for 8 parameters among node positive patients, and 4 parameters among node negative patients. The comparison between the Tata and Aarhus KBP model plans to manual plans from 67 institutions showed that the two KPBs had superior plan quality in 88-99% of instances.

CONCLUSION

KBP has the potential to generate high-quality plans across institutions and geo-ethnic populations by reducing inter-planner variation, thereby facilitating the global standardisation of radiotherapy for cervical cancer.

Dose prediction via deep learning to enhance treatment planning of lung radiotherapy including simultaneous integrated boost techniques

BACKGROUND

Recent studies have shown deep learning techniques are able to predict three-dimensional (3D) dose distributions of radiotherapy treatment plans. However, their use in dose prediction for treatments with varied prescription doses including simultaneous integrated boost (SIB), that is, using multiple prescription doses within the same plan, and benefit in improving plan quality should be validated.

PURPOSE

To investigate the feasibility and potential benefit of using deep learning to predict dose distribution of volumetric modulated arc therapy (VMAT) including SIB techniques and improve treatment planning for patients with lung cancer.

METHODS

The dose prediction model was trained with 93 retrospective clinical VMAT plans for patients with lung cancer from our institutional patient database. The prescription doses of these plans ranged from 35 to 72 Gy, with various fractionation schemes. We used a 3D U-Net architecture to predict 3D dose distributions with 75 plans for training and 18 plans for testing. Model input consisted of computed tomography (CT) images, target and normal tissue contours and prescription doses. We first evaluated model accuracy by comparing the predicted and clinical plan doses for the test set, and then performed replanning according to predicted dose distributions. Furthermore, we evaluated the model prospectively in an additional set of 10 patients from our institution by two approaches where dose prediction was either blinded or provided to treatment planners. We then assessed whether dose prediction could identify suboptimal plan quality and how it affects plan quality if adopted in clinical planning workflow.

RESULTS

The dose prediction model achieved good agreement between the predicted and clinical plan dose distributions, with a mean dose difference of -0.49 ± 0.54 Gy across the test set. The replanning study guided by dose prediction showed that a small subset of the original plans could benefit from improvements regarding sparing of the spinal cord and esophagus. The analysis of the prospective dataset, with initial and final clinical plans generated in the absence of dose prediction, showed that the predicted doses were able to identify possible improvements of target coverage and normal tissue sparing in the initial plans similar to those made by the final plans for majority of the patients, but in varied magnitudes. Moreover, the plans generated with dose prediction guidance were able to consistently improve normal tissue sparing compared to the plans generated without dose prediction guidance.

CONCLUSIONS

We demonstrated that our deep learning model can consistently predict high quality VMAT lung plans for a variety of prescription doses. The dose prediction tool was also effective in identifying suboptimal plan quality, suggesting its potential benefit in automated treatment planning and evaluation.

Enhancing automated right-sided early-stage breast cancer treatments via deep learning model adaptation without additional training

BACKGROUND

Input data curation and model training are essential, but time-consuming steps in building a deep-learning (DL) auto-planning model, ensuring high-quality data and optimized performance. Ideally, one would prefer a DL model that exhibits the same high-quality performance as a trained model without the necessity of undergoing such time-consuming processes. That goal can be achieved by providing models that have been trained on a given dataset and are capable of being fine-tuned for other ones, requiring no additional training.

PURPOSE

To streamline the process for producing an automated right-sided breast (RSB) treatment planning technique adapting a DL model originally trained on left-sided breast (LSB) patients via treatment planning system (TPS) specific tools only, thereby eliminating the need for additional training.

METHODS

The adaptation process involved the production of a predicted dose (PD) for the RSB by swapping from left-to-right the symmetric structures in association with the tuning of the initial LSB model settings for each of the two steps that follow the dose prediction: the predict settings for the post-processing of the PD (ppPD) and the mimic settings for the dose mimicking, respectively. Thirty patients were involved in the adaptation process: Ten manual plans were chosen as ground truth for tuning the LSB model settings, and the adapted RSB model was validated against 20 manual plans. During model tuning, PD, ppPD, and mimicked dose (MD) were iteratively compared to the manual dose according to the new RSB model settings configurations. For RSB model validation, only MD was involved in the planning comparison. Subsequently, the model was applied to 10 clinical patients. Manual and automated plans were compared using a site-specific list of dose-volume requirements.

RESULTS

PD for the RSB model required substantial corrections as it differed significantly from manual doses in terms of mean dose to the heart (+11.1 Gy) and right lung (+4.4 Gy), and maximum dose to the left lung (+6.4 Gy) and right coronary (+11.5 Gy). Such discrepancies were first addressed by producing a ppPD always superior to the manual dose by changing or introducing new predict settings. Second, the mimic settings were also reformulated to ensure a MD not inferior to the manual dose. The final adapted version of the RSB model settings, for which MD was found to be not significantly different than the manual dose except for a better right lung sparing (-1.1 Gy average dose), was retained for the model validation. In RSB model validation, a few significant-yet not clinically relevant-differences were noted, with the right lung being more spared in auto-plans (-0.6 Gy average dose) and the maximum dose to the left lung being lower in the manual plans (-0.8 Gy). The clinical plans returned dose distributions not significantly different than the validation plans.

CONCLUSION

The proposed technique adapts a DL model initially trained for LSB cancer for right-sided patients. It involves swapping the dose predictions from left to right and adjusting model settings, without the need for additional training. This technique-specific to a TPS-could be transposed to other TPS platforms.

A deep learning-based peer review method for radiotherapy planning

BACKGROUND

Quality control (QC) in radiotherapy planning is crucial for ensuring treatment efficacy and patient safety. Traditionally, QC relies on standard indicators and subjective assessments, which may lead to inconsistencies.

PURPOSE

This study aims to develop a novel peer review method for personalized QC in radiotherapy planning, which is based on patient anatomical information, and utilizes deep learning dose prediction and a statistical model.

METHODS

A UNet model was trained on 139 nasopharyngeal carcinoma patients to predict 3D dose distribution, with plans divided into 95 for training, 20 for validation, and 24 for testing. For the clinical evaluation (24 items in total) of organs at risk (OAR), the QC interval (qualified, acceptable, or unqualified) for these items was set according to the model accuracy. Peer review was performed on another 29 clinical treatment plans, the items identified by the model as requiring optimization and improvement were optimized, and the effectiveness of the peer review method was tested.

RESULTS

The predicted mean voxel-based dose difference was 0.29 ± 0.13 Gy. For most evaluation items, the model prediction results were comparable to the planned results. Peer review results suggested that 66% of the plans were acceptable or unqualified. After optimization, 100% of the acceptable plans and 47% of the unqualified plans became qualified, and 20% of the unqualified plans became acceptable.

CONCLUSIONS

A deep learning dose prediction model based on patient information can be used to develop personalized QC in radiotherapy planning and can help improve the quality of radiotherapy plans.

Knowledge-based planning for fully automated radiation therapy treatment planning of 10 different cancer sites

PURPOSE

Radiation treatment planning is highly complex and can have significant inter- and intra-planner inconsistency, as well as variability in planning time and plan quality. Knowledge-based planning (KBP) is a tool that can be used to efficiently produce high-quality, consistent, clinically acceptable plans, independent of planner skills and experience. In this study, we created and validated multiple clinically acceptable and fully automatable KBP models, with the goal of creating VMAT plans without user intervention.

METHODS

Ten KBP models were configured using high quality clinical plans from a single institution. They were then honed to be part of a fully automatable system by incorporating scriptable planning structures, plan creation, and plan optimization. These models were verified and validated using quantitative (model statistics) and qualitative (dose-volume histogram estimation review) analysis. The resulting KBP-generated plans were reviewed by physicians and rated for clinical acceptability.

RESULTS

Autoplanning models were created for anorectal, bladder, breast/chest wall, cervix, esophagus, head and neck, liver, lung/mediastinum, prostate, and prostate with nodes treatment sites. All models were successfully created to be part of a fully automated system without the need for human intervention to create a fully optimized plan. The physician review indicated that, on average, 88% of all KBP-generated plans were "acceptable as is" and 98% were "acceptable after minor edits."

CONCLUSION

KBP models for multiple treatment sites were used as a basis to generate fully automatable, efficient, consistent, high-quality, and clinically acceptable plans. These plans do not require human intervention, demonstrating the potential this work has to significantly impact treatment planning workflows.

Clinical commissioning and introduction of an in-house artificial intelligence (AI) platform for automated head and neck intensity modulated radiation therapy (IMRT) treatment planning

BACKGROUND AND PURPOSE

To describe the clinical commissioning of an in-house artificial intelligence (AI) treatment planning platform for head-and-neck (HN) Intensity Modulated Radiation Therapy (IMRT).

MATERIALS AND METHODS

The AI planning platform has three components: (1) a graphical user interface (GUI) is built within the framework of a commercial treatment planning system (TPS). The GUI allows AI models to run remotely on a designated workstation configured with GPU acceleration. (2) A template plan is automatically prepared involving both clinical and AI considerations, which include contour evaluation, isocenter placement, and beam/collimator jaw placement. (3) A well-orchestrated suite of AI models predicts optimal fluence maps, which are imported into TPS for dose calculation followed by an optional automatic fine-tuning. Six AI models provide flexible tradeoffs in parotid sparing and Planning Target Volume (PTV)-organ-at-risk (OAR) preferences. Planners could examine the plan dose distribution and make further modifications as clinically needed. The performance of the AI plans was compared to the corresponding clinical plans.

RESULTS

The average plan generation time including manual operations was 10-15  min per case, with each AI model prediction taking ∼1 s. The six AI plans form a wide range of tradeoff choices between left and right parotids and between PTV and OARs compared with corresponding clinical plans, which correctly reflected their tradeoff designs.

CONCLUSION

The in-house AI IMRT treatment planning platform was developed and is available for clinical use at our institution. The process demonstrates outstanding performance and robustness of the AI platform and provides sufficient validation.

Critical assessment of knowledge-based models for craniospinal irradiation of paediatric patients

Background and purpose

Knowledge-Based Planning (KBP) is increasingly used to standardize and optimize radiotherapy planning. This study aims to develop, refine, and compare multicentric KBP models for craniospinal irradiation (CSI) in pediatric patients.

Materials and methods

A total of 113 CSI treatments from three Italian centers were collected, comprising Computed Tomography scans, target and organ contours, and treatment plans. Treatment techniques included Helical Tomotherapy (HT) and Volumetric Modulated Arc Therapy (VMAT). Three KBP models were developed: a full model (F-model) using data from 87 patients, a reduced model (R-model) based on a subset of the same sample, and a replanned model (RP-model) using KBP re-optimized plans. Models' quality was evaluated using goodness-of-fit and goodness-of-prediction metrics, and their performance was assessed on a validation set of 26 patients through dose-volume histogram (DVH) comparisons, prediction bias, and variance analysis.

Results

The F-model and R-model exhibited similar quality and predictive ability, reflecting the variability of the original dataset and resulting in broad prediction intervals in low to mid-dose ranges. The RP-model achieved the highest quality, with narrower prediction bands. The RP-model is preferable for standardizing planning across centers, while the F-model is better suited for quality assurance as it captures clinical variability.

Conclusions

KBP models can effectively predict DVHs despite extreme geometric variability. However, models trained on highly variable datasets cannot simultaneously achieve high precision and accuracy. Comparing KBP models is essential for establishing benchmarks to meet specific clinical goals, particularly for complex pediatric CSI treatments.

Clinical Impact of Contouring Variability for Prostate Cancer Tumor Boost

PURPOSE

The focal radiation therapy (RT) boost technique was shown in a phase III randomized controlled trial (RCT) to improve prostate cancer outcomes without increasing toxicity. This technique relies on the accurate delineation of prostate tumors on MRI. A recent prospective study evaluated radiation oncologists' accuracy when asked to delineate prostate tumors on MRI and demonstrated high variability in tumor contours. We sought to evaluate the impact of contour variability and inaccuracy on predicted clinical outcomes. We hypothesized that radiation oncologists' contour inaccuracies would yield meaningfully worse clinical outcomes.

METHODS AND MATERIALS

Forty-five radiation oncologists and 2 expert radiologists contoured prostate tumors on 30 patient cases. Of these cases, those with CT simulation or diagnostic CT available were selected for analysis. A knowledge-based planning model was developed to generate focal RT boost plans for each contour per the RCT protocol. The probability of biochemical failure (BF) was determined using a model from the RCT. The primary metric evaluated was delta BF (DBF = Participant BF - Expert BF). An absolute increase in BF ≥5% was considered clinically meaningful.

RESULTS

Eight patient cases and 394 target volumes for focal RT boost planning were included in this analysis. In general, participant plans were associated with worse predicted clinical outcomes compared to the expert plan, with an average absolute increase in BF of 4.3%. Of participant plans, 37% were noted to have an absolute increase in BF of 5% or more.

CONCLUSIONS

Radiation oncologists' attempts to contour tumor targets for focal RT boost are frequently inaccurate enough to yield meaningfully inferior clinical outcomes for patients.

Improving 3D dose prediction for breast radiotherapy using novel glowing masks and gradient-weighted loss functions

BACKGROUND

The quality of treatment plans for breast cancer can vary greatly. This variation could be reduced by using dose prediction to automate treatment planning. Our work investigates novel methods for training deep-learning models that are capable of producing high-quality dose predictions for breast cancer treatment planning.

PURPOSE

The goal of this work was to compare the performance impact of two novel techniques for deep learning dose prediction models for tangent field treatments for breast cancer. The first technique, a "glowing" mask algorithm, encodes the distance from a contour into each voxel in a mask. The second, a gradient-weighted mean squared error (MSE) loss function, emphasizes the error in high-dose gradient regions in the predicted image.

METHODS

Four 3D U-Net deep learning models were trained using the planning CT and contours of the heart, lung, and tumor bed as inputs. The dataset consisted of 305 treatment plans split into 213/46/46 training/validation/test sets using a 70/15/15% split. We compared the impact of novel "glowing" anatomical mask inputs and a novel gradient-weighted MSE loss function to their standard counterparts, binary anatomical masks, and MSE loss, using an ablation study methodology. To assess performance, we examined the mean error and mean absolute error (ME/MAE) in dose across all within-body voxels, the error in mean dose to heart, ipsilateral lung, and tumor bed, dice similarity coefficient (DSC) across isodose volumes defined by 0%-100% prescribed dose thresholds, and gamma analysis (3%/3 mm).

RESULTS

The combination of novel glowing masks and gradient weighted loss function yielded the best-performing model in this study. This model resulted in a mean ME of 0.40%, MAE of 2.70%, an error in mean dose to heart and lung of -0.10 and 0.01 Gy, and an error in mean dose to the tumor bed of -0.01%. The median DSC at 50/95/100% isodose levels were 0.91/0.87/0.82. The mean 3D gamma pass rate (3%/3 mm) was 93%.

CONCLUSIONS

This study found the combination of novel anatomical mask inputs and loss function for dose prediction resulted in superior performance to their standard counterparts. These results have important implications for the field of radiotherapy dose prediction, as the methods used here can be easily incorporated into many other dose prediction models for other treatment sites. Additionally, this dose prediction model for breast radiotherapy has sufficient performance to be used in an automated planning pipeline for tangent field radiotherapy and has the major benefit of not requiring a PTV for accurate dose prediction.

Improvement of accumulated dose distribution in combined cervical cancer radiotherapy with deep learning-based dose prediction

Purpose

Difficulties remain in dose optimization and evaluation of cervical cancer radiotherapy that combines external beam radiotherapy (EBRT) and brachytherapy (BT). This study estimates and improves the accumulated dose distribution of EBRT and BT with deep learning-based dose prediction.

Materials and methods

A total of 30 patients treated with combined cervical cancer radiotherapy were enrolled in this study. The dose distributions of EBRT and BT plans were accumulated using commercial deformable image registration. A ResNet-101-based deep learning model was trained to predict pixel-wise dose distributions. To test the role of the predicted accumulated dose in clinic, each EBRT plan was designed using conventional method and then redesigned referencing the predicted accumulated dose distribution. Bladder and rectum dosimetric parameters and normal tissue complication probability (NTCP) values were calculated and compared between the conventional and redesigned accumulated doses.

Results

The redesigned accumulated doses showed a decrease in mean values of V50, V60, and D2cc for the bladder (-3.02%, -1.71%, and -1.19 Gy, respectively) and rectum (-4.82%, -1.97%, and -4.13 Gy, respectively). The mean NTCP values for the bladder and rectum were also decreased by 0.02‰ and 0.98%, respectively. All values had statistically significant differences (p < 0.01), except for the bladder D2cc (p = 0.112).

Conclusion

This study realized accumulated dose prediction for combined cervical cancer radiotherapy without knowing the BT dose. The predicted dose served as a reference for EBRT treatment planning, leading to a superior accumulated dose distribution and lower NTCP values.

Radiotherapy Plan Quality Assurance in NRG Oncology Trials for Brain and Head/Neck Cancers: An AI-Enhanced Knowledge-Based Approach

The quality of radiation therapy (RT) treatment plans directly affects the outcomes of clinical trials. KBP solutions have been utilized in RT plan quality assurance (QA). In this study, we evaluated the quality of RT plans for brain and head/neck cancers enrolled in multi-institutional clinical trials utilizing a KBP approach. The evaluation was conducted on 203 glioblastoma (GBM) patients enrolled in NRG-BN001 and 70 nasopharyngeal carcinoma (NPC) patients enrolled in NRG-HN001. For each trial, fifty high-quality photon plans were utilized to build a KBP photon model. A KBP proton model was generated using intensity-modulated proton therapy (IMPT) plans generated on 50 patients originally treated with photon RT. These models were then applied to generate KBP plans for the remaining patients, which were compared against the submitted plans for quality evaluation, including in terms of protocol compliance, target coverage, and organ-at-risk (OAR) doses. RT plans generated by the KBP models were demonstrated to have superior quality compared to the submitted plans. KBP IMPT plans can decrease the variation of proton plan quality and could possibly be used as a tool for developing improved plans in the future. Additionally, the KBP tool proved to be an effective instrument for RT plan QA in multi-center clinical trials.

On the necessity of specialized knowledge-based models for SBRT prostate treatments plans

PURPOSE

Test whether a well-grounded KBP model trained on moderately hypo-fractionated prostate treatments can be used to satisfactorily drive the optimization of SBRT prostate treatments.

MATERIALS AND METHODS

A KBP model (SBRT-model) was developed, trained and validated using the first forty-seven clinically treated VMAT SBRT prostate plans (42.7 Gy/7fx or 36.25 Gy/5fx). The performance and robustness of this model were compared against a high-quality KBP-model (ST-model) that was already clinically adopted for hypo-fractionated (70 Gy/28fx and 60 Gy/20fx) prostate treatments. The two models were compared in terms of their predictions robustness, and the quality of their outcomes were evaluated against a set of reference clinical SBRT plans. Plan quality was assessed using DVH metrics, blinded clinical ranking, and a dedicated Plan Quality Metric algorithm.

RESULTS

The plan libraries of the two models were found to share a high degree of anatomical similarity. The overall quality (APQM%) of the plans obtained both with the ST- and SBRT-models was compatible with that of the original clinical plans, namely (93.7 ± 4.1)% and (91.6 ± 3.9)% vs (92.8.9 ± 3.6)%. Plans obtained with the ST-model showed significantly higher target coverage (PTV V95%): (97.9 ± 0.8)% vs (97.1 ± 0.9)% (p < 0.05). Conversely, plans optimized following the SBRT-model showed a small but not-clinically relevant increase in OAR sparing. ST-model generally provided more reliable predictions than SBRT-model. Two radiation oncologists judged as equivalent the plans based on the KBP prediction, which was also judged better that reference clinical plans.

CONCLUSION

A KBP model trained on moderately fractionated prostate treatment plans provided optimal SBRT prostate plans, with similar or larger plan quality than an embryonic SBRT-model based on a limited number of cases.

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.

Deep learning-based synthetic dose-weighted LET map generation for intensity modulated proton therapy

The advantage of proton therapy as compared to photon therapy stems from the Bragg peak effect, which allows protons to deposit most of their energy directly at the tumor while sparing healthy tissue. However, even with such benefits, proton therapy does present certain challenges. The biological effectiveness differences between protons and photons are not fully incorporated into clinical treatment planning processes. In current clinical practice, the relative biological effectiveness (RBE) between protons and photons is set as constant 1.1. Numerous studies have suggested that the RBE of protons can exhibit significant variability. Given these findings, there is a substantial interest in refining proton therapy treatment planning to better account for the variable RBE. Dose-average linear energy transfer (LETd) is a key physical parameter for evaluating the RBE of proton therapy and aids in optimizing proton treatment plans. Calculating precise LETddistributions necessitates the use of intricate physical models and the execution of specialized Monte-Carlo simulation software, which is a computationally intensive and time-consuming progress. In response to these challenges, we propose a deep learning based framework designed to predict the LETddistribution map using the dose distribution map. This approach aims to simplify the process and increase the speed of LETdmap generation in clinical settings. The proposed CycleGAN model has demonstrated superior performance over other GAN-based models. The mean absolute error (MAE), peak signal-to-noise ratio and normalized cross correlation of the LETdmaps generated by the proposed method are 0.096 ± 0.019 keVμm-1, 24.203 ± 2.683 dB, and 0.997 ± 0.002, respectively. The MAE of the proposed method in the clinical target volume, bladder, and rectum are 0.193 ± 0.103, 0.277 ± 0.112, and 0.211 ± 0.086 keVμm-1, respectively. The proposed framework has demonstrated the feasibility of generating synthetic LETdmaps from dose maps and has the potential to improve proton therapy planning by providing accurate LETdinformation.

Benchmarking Automated Machine Learning-Enhanced Planning With Ethos Against Manual and Knowledge-Based Planning for Locally Advanced Lung Cancer

Purpose

Currently, there is insufficient guidance for standard fractionation lung planning using the Varian Ethos adaptive treatment planning system and its unique intelligent optimization engine. Here, we address this gap in knowledge by developing a methodology to automatically generate high-quality Ethos treatment plans for locally advanced lung cancer.

Methods and Materials

Fifty patients previously treated with manually generated Eclipse plans for inoperable stage IIIA-IIIC non-small cell lung cancer were included in this institutional review board-approved retrospective study. Fifteen patient plans were used to iteratively optimize a planning template for the Daily Adaptive vs Non-Adaptive External Beam Radiation Therapy With Concurrent Chemotherapy for Locally Advanced Non-Small Cell Lung Cancer: A Prospective Randomized Trial of an Individualized Approach for Toxicity Reduction (ARTIA-Lung); the remaining 35 patients were automatically replanned without intervention. Ethos plan quality was benchmarked against clinical plans and reoptimized knowledge-based RapidPlan (RP) plans, then judged using standard dose-volume histogram metrics, adherence to clinical trial objectives, and qualitative review.

Results

Given equal prescription target coverage, Ethos-generated plans showed improved primary and nodal planning target volume V95% coverage (P < .001) and reduced lung gross tumor volume V5 Gy and esophagus D0.03 cc metrics (P ≤ .003) but increased mean esophagus and brachial plexus D0.03 cc metrics (P < .001) compared with RP plans. Eighty percent, 49%, and 51% of Ethos, clinical, and RP plans, respectively, were "per protocol" or met "variation acceptable" ARTIA-Lung planning metrics. Three radiation oncologists qualitatively scored Ethos plans, and 78% of plans were clinically acceptable to all reviewing physicians, with no plans receiving scores requiring major changes.

Conclusions

A standard Ethos template produced lung radiation therapy plans with similar quality to RP plans, elucidating a viable approach for automated plan generation in the Ethos adaptive workspace.

Improving knowledge-based treatment planning for lung cancer radiotherapy with automatic multi-criteria optimized training plans

BACKGROUND

Knowledge-based planning (KBP) is a method for automated radiotherapy treatment planning where appropriate optimization objectives for new patients are predicted based on a library of training plans. KBP can save time and improve organ at-risk sparing and inter-patient consistency compared to manual planning, but its performance depends on the quality of the training plans. We used another system for automated planning, which generates multi-criteria optimized (MCO) plans based on a wish list, to create training plans for the KBP model, to allow seamless integration of knowledge from a new system into clinical routine. Model performance was compared for KBP models trained with manually created and automatic MCO treatment plans.

MATERIAL AND METHODS

Two RapidPlan models with the same 30 locally advanced non-small cell lung cancer patients included were created, one containing manually created clinical plans (RP_CLIN) and one containing fully automatic multi-criteria optimized plans (RP_MCO). For 15 validation patients, model performance was compared in terms of dose-volume parameters and normal tissue complication probabilities, and an oncologist performed a blind comparison of the clinical (CLIN), RP_CLIN, and RP_MCO plans.

RESULTS

The heart and esophagus doses were lower for RP_MCO compared to RP_CLIN, resulting in an average reduction in the risk of 2-year mortality by 0.9 percentage points and the risk of acute esophageal toxicity by 1.6 percentage points with RP_MCO. The oncologist preferred the RP_MCO plan for 8 patients and the CLIN plan for 7 patients, while the RP_CLIN plan was not preferred for any patients.

CONCLUSION

RP_MCO improved OAR sparing compared to RP_CLIN and was selected for implementation in the clinic. Training a KBP model with clinical plans may lead to suboptimal output plans, and making an extra effort to optimize the library plans in the KBP model creation phase can improve the plan quality for many future patients.

Automated evaluation for rapid implementation of knowledge-based radiotherapy planning models

PURPOSE

Knowledge-based planning (KBP) offers the ability to predict dose-volume metrics based on information extracted from previous plans, reducing plan variability and improving plan quality. As clinical integration of KBP is increasing there is a growing need for quantitative evaluation of KBP models. A .NET-based application, RapidCompare, was created for automated plan creation and analysis of Varian RapidPlan models.

METHODS

RapidCompare was designed to read calculation parameters and a list of reference plans. The tool copies the reference plan field geometry and structure set, applies the RapidPlan model, optimizes the KBP plan, and generates data for quantitative evaluation of dose-volume metrics. A cohort of 85 patients, divided into training (50), testing (10), and validation (25) groups, was used to demonstrate the utility of RapidCompare. After training and tuning, the KBP model was paired with three different optimization templates to compare various planning strategies in the validation cohort. All templates used the same set of constraints for the planning target volume (PTV). For organs-at-risk, the optimization template provided constraints using the whole dose-volume histogram (DVH), fixed-dose/volume points, or generalized equivalent uniform dose (gEUD). The resulting plans from each optimization approach were compared using DVH metrics.

RESULTS

RapidCompare allowed for the automated generation of 75 total plans for comparison with limited manual intervention. In comparing optimization techniques, the Dose/Volume and Lines optimization templates generated plans with similar DVH metrics, with a slight preference for the Lines technique with reductions in heart V30Gy and spinal cord max dose. The gEUD model produced high target heterogeneity.

CONCLUSION

Automated evaluation allowed for the exploration of multiple optimization templates in a larger validation cohort than would have been feasible using a manual approach. A final KBP model using line optimization objectives produced the highest quality plans without human intervention.

Deep learning-based dose map prediction for high-dose-rate brachytherapy

Background. Creating a clinically acceptable plan in the time-sensitive clinic workflow of brachytherapy is challenging. Deep learning-based dose prediction techniques have been reported as promising solutions with high efficiency and accuracy. However, current dose prediction studies mainly target EBRT which are inappropriate for brachytherapy, the model designed specifically for brachytherapy has not yet well-established.Purpose. To predict dose distribution in brachytherapy using a novel Squeeze and Excitation Attention Net (SE_AN) model.Method. We hypothesized the tracks of192Ir inside applicators are essential for brachytherapy dose prediction. To emphasize the applicator contribution, a novel SE module was integrated into a Cascaded UNet to recalibrate informative features and suppress less useful ones. The Cascaded UNet consists of two stacked UNets, with the first designed to predict coarse dose distribution and the second added for fine-tuning 250 cases including all typical clinical applicators were studied, including vaginal, tandem and ovoid, multi-channel, and free needle applicators. The developed SE_AN was subsequently compared to the classic UNet and classic Cascaded UNet (without SE module) models. The model performance was evaluated by comparing the predicted dose against the clinically approved plans using mean absolute error (MAE) of DVH metrics, includingD2ccandD90%.Results. The MAEs of DVH metrics demonstrated that SE_AN accurately predicted the dose with 0.37 ± 0.25 difference for HRCTVD90%, 0.23 ± 0.14 difference for bladderD2cc, and 0.28 ± 0.20 difference for rectumD2cc. In comparison studies, UNet achieved 0.34 ± 0.24 for HRCTV, 0.25 ± 0.20 for bladder, 0.25 ± 0.21 for rectum, and Cascaded UNet achieved 0.42 ± 0.31 for HRCTV, 0.24 ± 0.19 for bladder, 0.23 ± 0.19 for rectum.Conclusion. We successfully developed a method specifically for 3D brachytherapy dose prediction. Our model demonstrated comparable performance to clinical plans generated by experienced dosimetrists. The developed technique is expected to improve the standardization and quality control of brachytherapy treatment planning.

Online adaptive planning methods for intensity-modulated radiotherapy

Online adaptive radiation therapy aims at adapting a patient's treatment plan to their current anatomy to account for inter-fraction variations before daily treatment delivery. As this process needs to be accomplished while the patient is immobilized on the treatment couch, it requires time-efficient adaptive planning methods to generate a quality daily treatment plan rapidly. The conventional planning methods do not meet the time requirement of online adaptive radiation therapy because they often involve excessive human intervention, significantly prolonging the planning phase. This article reviews the planning strategies employed by current commercial online adaptive radiation therapy systems, research on online adaptive planning, and artificial intelligence's potential application to online adaptive planning.

Updating a clinical Knowledge-Based Planning prediction model for prostate radiotherapy

BACKGROUND AND PURPOSE

Clinical knowledge-based planning (KBP) models dedicated to prostate radiotherapy treatment may require periodical updates to remain relevant and to adapt to possible changes in the clinic. This study proposes a paired comparison of two different update approaches through a longitudinal analysis.

MATERIALS AND METHODS

A clinically validated KBP model for moderately hypofractionated prostate therapy was periodically updated using two approaches: one was targeted at achieving the biggest library size (M_t), while the other one at achieving the highest mean sample quality (R_t). Four subsequent updates were accomplished. The goodness, robustness and quality of the outcomes were measured and compared to those of the common ancestor. Plan quality was assessed through the Plan Quality Metric (PQM) and plan complexity was monitored.

RESULTS

Both update procedures allowed for an increase in the OARs sparing between +3.9 % and +19.2 % compared to plans generated by a human planner. Target coverage and homogeneity slightly reduced [-0.2 %;-14.7 %] while plan complexity showed only minor changes. Increasing the sample size resulted in more reliable predictions and improved goodness-of-fit, while increasing the mean sample quality improved the outcomes but slightly reduced the models reliability.

CONCLUSIONS

Repeated updates of clinical KBP models can enhance their robustness, reliability and the overall quality of automatically generated plans. The periodical expansion of the model sample accompanied by the removal of the unacceptable low quality plans should maximize the benefits of the updates while limiting the associated workload.

Knowledge-Based Quality Assurance and Model Maintenance in Lung Cancer Radiation Therapy in a Statewide Quality Consortium of Academic and Community Practice Centers

PURPOSE

Locally advanced lung cancer (LALC) treatment planning is often complex due to challenging tradeoffs related to large targets near organs at risk, making the judgment of plan quality difficult. The purpose of this work was to update and maintain a multi-institutional knowledge-based planning (KBP) model developed by a statewide consortium of academic and community practices for use as a plan quality assurance (QA) tool.

METHODS AND MATERIALS

Sixty LALC volumetric-modulated arc therapy plans from 2021 were collected from 24 institutions. Plan quality was scored, with high-quality clinical (HQC) plans selected to update a KBP model originally developed in 2017. The model was validated via automated KBP planning, with 20 cases excluded from the model. Differences in dose-volume histogram metrics in the clinical plans, 2017 KBP model plans, and 2022 KBP model plans were compared. Twenty recent clinical cases not meeting consortium quality metrics were replanned with the 2022 model to investigate potential plan quality improvements.

RESULTS

Forty-seven plans were included in the final KBP model. Compared with the clinical plans, the 2022 model validation plans improved 60%, 65%, and 65% of the lung V20Gy, mean heart dose, and spinal canal D0.03cc metrics, respectively. The 2022 model showed improvements from the 2017 model in hot spot management at the cost of greater lung doses. Of the 20 recent cases not meeting quality metrics, 40% of the KBP model-replanned cases resulted in acceptable plans, suggesting potential clinical plan improvements.

CONCLUSIONS

A multi-institutional KBP model was updated using plans from a statewide consortium. Multidisciplinary plan review resulted in HQC model training plans and model validation resulted in acceptable quality plans. The model proved to be effective at identifying potential plan quality improvements. Work is ongoing to develop web-based training plan review tools and vendor-agnostic platforms to provide the model as a QA tool statewide.

Leveraging intelligent optimization for automated, cardiac-sparing accelerated partial breast treatment planning

Background

Accelerated partial breast irradiation (APBI) yields similar rates of recurrence and cosmetic outcomes as compared to whole breast radiation therapy (RT) when patients and treatment techniques are appropriately selected. APBI combined with stereotactic body radiation therapy (SBRT) is a promising technique for precisely delivering high levels of radiation while avoiding uninvolved breast tissue. Here we investigate the feasibility of automatically generating high quality APBI plans in the Ethos adaptive workspace with a specific emphasis on sparing the heart.

Methods

Nine patients (10 target volumes) were utilized to iteratively tune an Ethos APBI planning template for automatic plan generation. Twenty patients previously treated on a TrueBeam Edge accelerator were then automatically replanned using this template without manual intervention or reoptimization. The unbiased validation cohort Ethos plans were benchmarked via adherence to planning objectives, a comparison of DVH and quality indices against the clinical Edge plans, and qualitative reviews by two board-certified radiation oncologists.

Results

85% (17/20) of automated validation cohort plans met all planning objectives; three plans did not achieve the contralateral lung V1.5Gy objective, but all other objectives were achieved. Compared to the Eclipse generated plans, the proposed Ethos template generated plans with greater evaluation planning target volume (PTV_Eval) V100% coverage (p = 0.01), significantly decreased heart V1.5Gy (p< 0.001), and increased contralateral breast V5Gy, skin D0.01cc, and RTOG conformity index (p = 0.03, p = 0.03, and p = 0.01, respectively). However, only the reduction in heart dose was significant after correcting for multiple testing. Physicist-selected plans were deemed clinically acceptable without modification for 75% and 90% of plans by physicians A and B, respectively. Physicians A and B scored at least one automatically generated plan as clinically acceptable for 100% and 95% of planning intents, respectively.

Conclusions

Standard left- and right-sided planning templates automatically generated APBI plans of comparable quality to manually generated plans treated on a stereotactic linear accelerator, with a significant reduction in heart dose compared to Eclipse generated plans. The methods presented in this work elucidate an approach for generating automated, cardiac-sparing APBI treatment plans for daily adaptive RT with high efficiency.

Three-dimensional dose and LETD prediction in proton therapy using artificial neural networks

PURPOSE

Challenges in proton therapy include identifying patients most likely to benefit; ensuring consistent, high-quality plans as its adoption becomes more widespread; and recognizing biological uncertainties that may be related to increased relative biologic effectiveness driven by linear energy transfer (LET). Knowledge-based planning (KBP) is a domain that may help to address all three.

METHODS

Artificial neural networks were trained using 117 unique treatment plans and associated dose and dose-weighted LET (LETD ) distributions. The data set was split into training (n = 82), validation (n = 17), and test (n = 18) sets. Model performance was evaluated on the test set using dose- and LETD -volume metrics in the clinical target volume (CTV) and nearby organs at risk and Dice similarity coefficients (DSC) comparing predicted and planned isodose lines at 50%, 75%, and 95% of the prescription dose.

RESULTS

Dose-volume metrics significantly differed (α = 0.05) between predicted and planned dose distributions in only one dose-volume metric, D2% to the CTV. The maximum observed root mean square (RMS) difference between corresponding metrics was 4.3 GyRBE (8% of prescription) for D1cc to optic chiasm. DSC were 0.90, 0.93, and 0.88 for the 50%, 75%, and 95% isodose lines, respectively. LETD -volume metrics significantly differed in all but one metric, L0.1cc of the brainstem. The maximum observed difference in RMS differences for LETD metrics was 1.0 keV/μm for L0.1cc to brainstem.

CONCLUSIONS

We have devised the first three-dimensional dose and LETD -prediction model for cranial proton radiation therapy has been developed. Dose accuracy compared favorably with that of previously published models in other treatment sites. The agreement in LETD supports future investigations with biological doses in mind to enable the full potential of KBP in proton therapy.

Knowledge-based planning using both the predicted DVH of organ-at risk and planning target volume

PURPOSE

The purpose of this study was to evaluate the performance of a knowledge-based planning (KBP) method in nasopharyngeal cancer radiotherapy using the predicted dose-volume histogram (DVH) of organ-at risk (OAR) and planning target volume (PTV).

METHODS AND MATERIALS

A total of 85 patients previously treated for nasopharyngeal cancer using 9-field 6-MV intensity-modulated radiation therapy (IMRT) were identified for training and 30 similar patients were identified for testing. The dosimetric deposition information, individual dose-volume histograms (IDVHs) induced by a series of fields with uniform-intensity irradiation, was used to predict both OAR and PTV DVH. Two KBP methods (KBP_OAR and KBP_OAR+PTV) were established for plan generation based on the DVH prediction. The KBP_OAR method utilized the dose constraints based on the predicted OAR DVH and the PTV dose constraints obtained according to the planning experience, while the KBP_OAR+PTV method applied the dose constraints based on the predicted OAR and PTV DVH. For the plan evaluation, the PTV dose coverage was used D₉₈ and D₂, and the maximum dose, mean dose or dose-volume parameters were used for the OARs. Statistical differences of the two KBP methods were tested with the Wilcoxon signed rank test.

RESULTS

For patients with T3 tumors, there was no significant difference between the KBP_OAR and KBP_OAR+PTV methods in dosimetric results at most OARs and PTVs. Both KBP methods achieved a similar number of plans meeting the dose requirements. For patients with T4 tumors, KBP_OAR+PTV reduced the maximum dose by more than 1 Gy in the body, spinal cord, optic nerve, eye and temporal lobes and reduced the V₅₀ value by more than 3.9% in the larynx and tongue without reducing the PTV dose compared with KBP_OAR. The KBP_OAR+PTV method increased the plans by more than 14.2% in meeting the maximum dose requirements at the body, optic nerve, mandible and eye and increased the plans by more than 21.4% in meeting the V₅₀ of the larynx and V₅₀ of the tongue when compared with the KBP_OAR method.

CONCLUSIONS

For patients with T3 tumors, no significant difference was found between the KBP_OAR and KBP_OAR+PTV methods in dosimetric results at most OARs and PTVs. For patients with T4 tumors, the KBP_OAR+PTV method performs better than the KBP_OAR method in improving the quality of the plans. Compared with the KBP_OAR method, dose sparing of some OARs was achieved without reducing PTV dose coverage and helped to increase the number of plans meeting the dose requirements when the KBP_OAR+PTV method was utilized.

Comprehensive dosimetric and clinical evaluation of lexicographic optimization-based planning for cervical cancer

AIM

In this study, a not yet commercially available fully-automated lexicographic optimization (LO) planning algorithm, called mCycle (Elekta AB, Stockholm, Sweden), was validated for cervical cancer.

MATERIAL AND METHODS

Twenty-four mono-institutional consecutive treatment plans (50 Gy/25 fx) delivered between November 2019 and April 2022 were retrospectively selected. The automatic re-planning was performed by mCycle, implemented in the Monaco TPS research version (v5.59.13), in which the LO and Multicriterial Optimization (MCO) are coupled with Monte Carlo calculation. mCycle optimization follows an a priori assigned priority list, the so-called Wish List (WL), representing a dialogue between the radiation oncologist and the planner, setting hard constraints and following objectives. The WL was tuned on a patient subset according to the institution's clinical protocol to obtain an optimal plan in a single optimization. This robust WL was then used to automatically re-plan the remaining patients. Manual plans (MP) and mCycle plans (mCP) were compared in terms of dose distributions, complexity (modulation complexity score, MCS), and delivery accuracy (perpendicular diode matrices, gamma analysis-passing ratio, PR). Their clinical acceptability was assessed through the blind choice of two radiation oncologists. Finally, a global quality score index (SI) was defined to gather into a single number the plan evaluation process.

RESULTS

The WL tuning requested four patients. The 20 automated re-planning tasks took three working days. The median optimization and calculation time can be estimated at 4 h and just over 1 h per MP and mCP, respectively. The dose comparison showed a comparable organ-at-risk spare. The planning target volume coverage increased (V95%: MP 98.0% [95.6-99.3]; mCP 99.2%[89.7-99.9], p >0.05). A significant increase has been registered in MCS (MP 0.29 [0.24-0.34]; mCP 0.26 [0.23-0.30], p <0.05) without affecting delivery accuracy (PR (3%/3mm): MP 97.0% [92.7-99.2]; mCP 97.1% [95.0-98.6], p >0.05). In the blind choice, all mCP results were clinically acceptable and chosen over MP in more than 75% of cases. The median SI score was 0.69 [0.41-0.84] and 0.73 [0.51-0.82] for MP and mCP, respectively (p >0.05).

CONCLUSIONS

mCycle plans were comparable to clinical manual plans, more complex but accurately deliverable and registering a similar SI. Automated plans outperformed manual plans in blinded clinical choice.

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.

A personalized DVH prediction model for HDR brachytherapy in cervical cancer treatment

Purpose

Although the knowledge-based dose-volume histogram (DVH) prediction has been largely researched and applied in External Beam Radiation Therapy, it is still less investigated in the domain of brachytherapy. The purpose of this study is to develop a reliable DVH prediction method for high-dose-rate brachytherapy plans.

Method

A DVH prediction workflow combining kernel density estimation (KDE), k-nearest neighbor (kNN), and principal component analysis (PCA) was proposed. PCA and kNN were first employed together to select similar patients based on principal component directions. 79 cervical cancer patients with different applicators inserted was included in this study. The KDE model was built based on the relationship between distance-to-target (DTH) and the dose in selected cases, which can be subsequently used to estimate the dose probability distribution in the validation set. Model performance of bladder and rectum was quantified by |ΔD_2cc|, |ΔD_1cc|, |ΔD_0.1cc|, |ΔD_max|, and |ΔD_mean| in the form of mean and standard deviation. The model performance between KDE only and the combination of kNN, PCA, and KDE was compared.

Result

20, 30 patients were selected for rectum and bladder based on KNN and PCA, respectively. The absolute residual between the actual plans and the predicted plans were 0.38 ± 0.29, 0.4 ± 0.32, 0.43 ± 0.36, 0.97 ± 0.66, and 0.13 ± 0.99 for |ΔD_2cc|, |ΔD_1cc|, |ΔD_0.1cc|, |ΔD_max|, and |ΔD_mean| in the bladder, respectively. For rectum, the corresponding results were 0.34 ± 0.27, 0.38 ± 0.33, 0.63 ± 0.57, 1.41 ± 0.99 and 0.23 ± 0.17, respectively. The combination of kNN, PCA, and KDE showed a significantly better prediction performance than KDE only, with an improvement of 30.3% for the bladder and 33.3% for the rectum.

Conclusion

In this study, a knowledge-based machine learning model was proposed and verified to accurately predict the DVH for new patients. This model is proved to be effective in our testing group in the workflow of HDR brachytherapy.

Clinical acceptability of fully automated external beam radiotherapy for cervical cancer with three different beam delivery techniques

Purpose

To fully automate CT‐based cervical cancer radiotherapy by automating contouring and planning for three different treatment techniques.

Methods

We automated three different radiotherapy planning techniques for locally advanced cervical cancer: 2D 4‐field‐box (4‐field‐box), 3D conformal radiotherapy (3D‐CRT), and volumetric modulated arc therapy (VMAT). These auto‐planning algorithms were combined with a previously developed auto‐contouring system. To improve the quality of the 4‐field‐box and 3D‐CRT plans, we used an in‐house, field‐in‐field (FIF) automation program. Thirty‐five plans were generated for each technique on CT scans from multiple institutions and evaluated by five experienced radiation oncologists from three different countries. Every plan was reviewed by two of the five radiation oncologists and scored using a 5‐point Likert scale.

Results

Overall, 87%, 99%, and 94% of the automatically generated plans were found to be clinically acceptable without modification for the 4‐field‐box, 3D‐CRT, and VMAT plans, respectively. Some customizations of the FIF configuration were necessary on the basis of radiation oncologist preference. Additionally, in some cases, it was necessary to renormalize the plan after it was generated to satisfy radiation oncologist preference.

Conclusion

Approximately, 90% of the automatically generated plans were clinically acceptable for all three planning techniques. This fully automated planning system has been implemented into the radiation planning assistant for further testing in resource‐constrained radiotherapy departments in low‐ and middle‐income countries.

Knowledge-based multi-institution plan prediction of whole breast irradiation with tangential fields

PURPOSE

To quantify inter-institute variability of Knowledge-Based (KB) models for right breast cancer patients treated with tangential fields whole breast irradiation (WBI).

MATERIALS AND METHODS

Ten institutions set KB models by using RapidPlan (Varian Inc.), following previously shared methodologies. Models were tested on 20 new patients from the same institutes, exporting DVH predictions of heart, ipsilateral lung, contralateral lung, and contralateral breast. Inter-institute variability was quantified by the inter-institute SD_int of predicted DVHs/Dmean. Association between lung sparing vs PTV coverage strategy was also investigated. The transferability of models was evaluated by the overlap of each model's geometric Principal Component (PC1) when applied to the test patients of the other 9 institutes.

RESULTS

The overall inter-institute variability of DVH/Dmean ipsilateral lung dose prediction, was less than 2% (20%-80% dose range) and 0.55 Gy respectively (1SD) for a 40 Gy in 15 fraction schedule; it was < 0.2 Gy for other OARs. Institute 6 showed the lowest mean dose prediction value and no overlap between PTV and ipsilateral lung. Once excluded, the predicted ipsilateral lung Dmean was correlated with median PTV D99% (R² = 0.78). PC1 values were always within the range of applicability (90th percentile) for 7 models: for 2 models they were outside in 1/18 cases. For the model of institute 6, it failed in 7/18 cases. The impact of inter-institute variability of dose calculation was tested and found to be almost negligible.

CONCLUSIONS

Results show limited inter-institute variability of plan prediction models translating in high inter-institute interchangeability, except for one of ten institutes. These results encourage future investigations in generating benchmarks for plan prediction incorporating inter-institute variability.

A practical method to improve the performance of knowledge-based VMAT planning for endometrial and cervical cancer

PURPOSE

The aim of this work was to demonstrate a practical and effective method to improve the performance of RapidPlan (RP) model.

METHODS

203 consecutive clinical VMAT plans (P₀) for cervical and endometrial cancer were used to train an RP model (M₀). The plans were then reoptimized by M₀ to generate 203 new plans (P₁). Compared with P₀, 150 plans with a lower mean dose (MD) of bladder, rectum and PBM were selected from P₁ to configure a new RP model (M₁). A final RP model (M₂) was trained using plans in M₁ and the remaining 53 plans from P₁ (excluding OARs with worse MD) and the corresponding plans from P₀ (only including OARs with better MD). The models were validated on the mentioned 53 plans (closed-loop set) and 46 patient cohorts outside the training library (open-loop set). p < 0.05 was considered statistically significant.

RESULTS

For closed-loop validation, the difference of D_2%, D_98% and CI_95% between groups was of no statistical significance, the homogeneity index (HI) was lower in the groups of RP models (p < 0.05). The MD of all OARs decreased monotonically in the sequence of the clinical group, group M₀, M₁ and M₂, except the MD of bowel in M₁ and MD of LFH in M₂. Similarly, for open-loop validation, there was no significant difference in D_2%, D_98% and HI between groups, but CI_95% was larger in the clinical group (p < 0.05). The MD of all OARs decreased monotonically in the sequence of the clinical group, group M₀, M₁ and M₂, with the exception of bowel in M₁.

CONCLUSION

The practical method of incorporating plan data of better-sparing OARs from both the clinical VMAT plans and the re-optimized plans could further improve the performance of the RP model.

Assessing the practicality of using a single knowledge‐based planning model for multiple linac vendors

Purpose

Knowledge‐based planning (KBP) has been shown to be an effective tool in quality control for intensity‐modulated radiation therapy treatment planning and generating high‐quality plans. Previous studies have evaluated its ability to create consistent plans across institutions and between planners within the same institution as well as its use as teaching tool for inexperienced planners. This study evaluates whether planning quality is consistent when using a KBP model to plan across different treatment machines.

Materials and methods

This study used a RapidPlan model (Varian Medical Systems) provided by the vendor, to which we added additional planning objectives, maximum dose limits, and planning structures, such that a clinically acceptable plan is achieved in a single optimization. This model was used to generate and optimize volumetric‐modulated arc therapy plans for a cohort of 50 patients treated for head‐neck cancer. Plans were generated using the following treatment machines: Varian 2100, Elekta Versa HD, and Varian Halcyon. A noninferiority testing methodology was used to evaluate the hypothesis that normal and target metrics in our autoplans were no worse than a set of clinically‐acceptable baseline plans by a margin of 1.8 Gy or 3% dose‐volume. The quality of these plans were also compared through the use of common clinical dose‐volume histogram criteria.

Results

The Versa HD met our noninferiority criteria for 23 of 34 normal and target metrics; while the Halcyon and Varian 2100 machines met our criteria for 24 of 34 and 26 of 34 metrics, respectively. The experimental plans tended to have less volume coverage for prescription dose planning target volume and larger hotspot volumes. However, comparable plans were generated across different treatment machines.

Conclusions

These results support the use of a head‐neck RapidPlan models in centralized planning workflows that support clinics with different linac models/vendors, although some fine‐tuning for targets may be necessary.

Evaluation of an automated template-based treatment planning system for radiotherapy of anal, rectal and prostate cancer

•

Automated treatment planning system compared to manual planning.

•

Equivalent plan quality between VMAT manually generated- and IMRT automatically generated plans.

•

Evaluation of anal, prostate and rectum treatment plans.

•

Generation of highly consistent IMRT automated plan within 2 to 3.5 min.

Background and purpose

The Ethos system has enabled online adaptive radiotherapy (oART) by implementing an automated treatment planning system (aTPS) for both intensity-modulated radiotherapy (IMRT) and volumetric modulated arc radiotherapy (VMAT) plan creation. The purpose of this study is to evaluate the quality of aTPS plans in the pelvic region.

Material and Methods

Sixty patients with anal (n = 20), rectal (n = 20) or prostate (n = 20) cancer were retrospectively re-planned with the aTPS. Three IMRT (7-, 9- and 12-field) and two VMAT (2 and 3 arc) automatically generated plans (APs) were created per patient. The duration of the automated plan generation was registered. The best IMRT-AP and VMAT-AP for each patient were selected based on target coverage and dose to organs at risk (OARs). The AP quality was analyzed and compared to corresponding clinically accepted and manually generated VMAT plans (MPs) using several clinically relevant dose metrics. Calculation-based pre-treatment plan quality assurance (QA) was performed for all plans.

Results

The median total duration to generate the five APs with the aTPS was 55 min, 39 min and 35 min for anal, prostate and rectal plans, respectively. The target coverage and the OAR sparing were equivalent for IMRT-APs and VMAT-MPs, while VMAT-Aps.

demonstrated lower target dose homogeneity and higher dose to some OARs. Both conformity and homogeneity index were equivalent (rectal) or better (anal and prostate) for IMRT-APs compared to VMAT-MPs. All plans passed the patient-specific QA tolerance limit.

Conclusions

The aTPS generates plans comparable to MPs within a short time-frame which is highly relevant for oART treatments.

Toward automatic beam angle selection for pencil-beam scanning proton liver Treatments: A deep learning-based approach

BACKGROUND

Dose deposition characteristics of proton radiation can be advantageous over photons. Proton treatment planning however poses additional challenges for the planners. Proton therapy is usually delivered with only a small number of beam angles, and the quality of a proton treatment plan is largely determined by the beam angles employed. Finding the optimal beam angles for a proton treatment plan requires time and experience, motivating the investigation of automatic beam angle selection methods.

PURPOSE

A deep learning-based approach to automatic beam angle selection is proposed for proton pencil-beam scanning treatment planning of liver lesions.

METHODS

We cast beam-angle selection as a multi-label classification problem. To account for angular boundary discontinuity, the underlying convolution neural network is trained with the proposed Circular Earth Mover's Distance based regularization and multi-label circular-smooth label technique. Furthermore, an analytical algorithm emulating proton treatment planners' clinical practice is employed in post-processing to improve the output of the model. Forty-nine patients that received proton liver treatments between 2017 and 2020 were randomly divided into training (n = 31), validation (n = 7), and test sets (n = 11). AI-selected beam angles were compared with those angles selected by human planners, and the dosimetric outcome was investigated by creating plans using knowledge-based treatment planning.

RESULTS

For 7 of the 11 cases in the test set, AI-selected beam angles agreed with those chosen by human planners to within 20 degrees (median angle difference = 10°; mean = 18.6°). Moreover, out of the total 22 beam angles predicted by the model, 15 (68%) were within 10 degrees of the human-selected angles. The high correlation in beam angles resulted in comparable dosimetric statistics between proton treatment plans generated using AI- and human-selected angles. For the cases with beam angle differences exceeding 20°, the dosimetric analysis showed similar plan quality although with different emphases on organ-at-risk sparing.

CONCLUSIONS

This pilot study demonstrated the feasibility of a novel deep learning-based beam angle selection technique. Testing on liver cancer patients showed that the resulting plans were clinically viable with comparable dosimetric quality to those using human-selected beam angles. In tandem with auto-contouring and knowledge-based treatment planning tools, the proposed model could represent a pathway for nearly fully automated treatment planning in proton therapy. This article is protected by copyright. All rights reserved.

Knowledge-based planning for the radiation therapy treatment plan quality assurance for patients with head and neck cancer

This study aimed to investigate the feasibility of using a knowledge‐based planning technique to detect poor quality VMAT plans for patients with head and neck cancer. We created two dose–volume histogram (DVH) prediction models using a commercial knowledge‐based planning system (RapidPlan, Varian Medical Systems, Palo Alto, CA) from plans generated by manual planning (MP) and automated planning (AP) approaches. DVHs were predicted for evaluation cohort 1 (EC1) of 25 patients and compared with achieved DVHs of MP and AP plans to evaluate prediction accuracy. Additionally, we predicted DVHs for evaluation cohort 2 (EC2) of 25 patients for which we intentionally generated plans with suboptimal normal tissue sparing while satisfying dose–volume limits of standard practice. Three radiation oncologists reviewed these plans without seeing the DVH predictions. We found that predicted DVH ranges (upper–lower predictions) were consistently wider for the MP model than for the AP model for all normal structures. The average ranges of mean dose predictions among all structures was 9.7 Gy (MP model) and 3.4 Gy (AP model) for EC1 patients. RapidPlan models identified 7 MP plans as outliers according to mean dose or D1% for at least one structure, while none of AP plans were flagged. For EC2 patients, 22 suboptimal plans were identified by prediction. While re‐generated AP plans validated that these suboptimal plans could be improved, 40 out of 45 structures with predicted poor sparing were also identified by oncologist reviews as requiring additional planning to improve sparing in the clinical setting. Our study shows that knowledge‐based DVH prediction models can be sufficiently accurate for plan quality assurance purposes. A prediction model built by a small cohort automatically‐generated plans was effective in detecting suboptimal plans. Such tools have potential to assist the plan quality assurance workflow for individual patients in the clinic.

A pilot study of machine-learning based automated planning for primary brain tumours

Purpose

High-quality radiotherapy (RT) planning for children and young adults with primary brain tumours is essential to minimize the risk of late treatment effects. The feasibility of using automated machine-learning (ML) to aid RT planning in this population has not previously been studied.

Methods and materials

We developed a ML model that identifies learned relationships between image features and expected dose in a training set of 95 patients with a primary brain tumour treated with focal radiotherapy to a dose of 54 Gy in 30 fractions. This ML method was then used to create predicted dose distributions for 15 previously-treated brain tumour patients across two institutions, as a testing set. Dosimetry to target volumes and organs-at-risk (OARs) were compared between the clinically-delivered (human-generated) plans versus the ML plans.

Results

The ML method was able to create deliverable plans in all 15 patients in the testing set. All ML plans were generated within 30 min of initiating planning. Planning target volume coverage with 95% of the prescription dose was attained in all plans. OAR doses were similar across most structures evaluated; mean doses to brain and left temporal lobe were lower in ML plans than manual plans (mean difference to left temporal, – 2.3 Gy, p = 0.006; mean differences to brain, – 1.3 Gy, p = 0.017), whereas mean doses to right cochlea and lenses were higher in ML plans (+ 1.6–2.2 Gy, p < 0.05 for each).

Conclusions

Use of an automated ML method to aid RT planning for children and young adults with primary brain tumours is dosimetrically feasible and can be successfully used to create high-quality 54 Gy RT plans. Further evaluation after clinical implementation is planned.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13014-021-01967-3.

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.

Assessment of Knowledge-Based Planning for Prostate Intensity Modulated Proton Therapy

Purpose

To assess the performance of a proton-specific knowledge based planning (KBPP) model in creation of robustly optimized intensity-modulated proton therapy (IMPT) plans for treatment of patients with prostate cancer.

Materials and Methods

Forty-five patients with localized prostate cancer, who had previously been treated with volumetric modulated arc therapy, were selected and replanned with robustly optimized IMPT. A KBPP model was generated from the results of 30 of the patients, and the remaining 15 patient results were used for validation. The KBPP model quality and accuracy were evaluated with the model-provided organ-at-risk regression plots and metrics. The KBPP quality was also assessed through comparison of expert and KBPP-generated IMPT plans for target coverage and organ-at-risk sparing.

Results

The resulting R ² (mean ± SD, 0.87 ± 0.07) between dosimetric and geometric features, as well as the χ² test (1.17 ± 0.07) between the original and estimated data, showed the model had good quality. All the KBPP plans were clinically acceptable. Compared with the expert plans, the KBPP plans had marginally higher dose-volume indices for the rectum V65Gy (0.8% ± 2.94%), but delivered a lower dose to the bladder (-1.06% ± 2.9% for bladder V65Gy). In addition, KBPP plans achieved lower hotspot (-0.67Gy ± 2.17Gy) and lower integral dose (-0.09Gy ± 0.3Gy) than the expert plans did. Moreover, the KBPP generated better plans that demonstrated slightly greater clinical target volume V95 (0.1% ± 0.68%) and lower homogeneity index (-1.13 ± 2.34).

Conclusions

The results demonstrated that robustly optimized IMPT plans created by the KBPP model are of high quality and are comparable to expert plans. Furthermore, the KBPP model can generate more-robust and more-homogenous plans compared with those of expert plans. More studies need to be done for the validation of the proton KBPP model at more-complicated treatment sites.

Knowledge-Based Planning for Robustly Optimized Intensity-Modulated Proton Therapy of Head and Neck Cancer Patients

PURPOSE

To assess the performance of a proton-specific knowledge-based planning (KBP) model in the creation of robustly optimized intensity-modulated proton therapy (IMPT) plans for treatment of advanced head and neck (HN) cancer patients.

METHODS

Seventy-three patients diagnosed with advanced HN cancer previously treated with volumetric modulated arc therapy (VMAT) were selected and replanned with robustly optimized IMPT. A proton-specific KBP model, RapidPlanPT (RPP), was generated using 53 patients (20 unilateral cases and 33 bilateral cases). The remaining 20 patients (10 unilateral and 10 bilateral cases) were used for model validation. The model was validated by comparing the target coverage and organ at risk (OAR) sparing in the RPP-generated IMPT plans with those in the expert plans. To account for the robustness of the plan, all uncertainty scenarios were included in the analysis.

RESULTS

All the RPP plans generated were clinically acceptable. For unilateral cases, RPP plans had higher CTV_primary V100 (1.59% ± 1.24%) but higher homogeneity index (HI) (0.7 ± 0.73) than had the expert plans. In addition, the RPP plans had better ipsilateral cochlea Dmean (-5.76 ± 6.11 Gy), with marginal to no significant difference between RPP plans and expert plans for all other OAR dosimetric indices. For the bilateral cases, the V100 for all clinical target volumes (CTVs) was higher for the RPP plans than for the expert plans, especially the CTV_primary V100 (5.08% ± 3.02%), with no significant difference in the HI. With respect to OAR sparing, RPP plans had a lower spinal cord Dmax (-5.74 ± 5.72 Gy), lower cochlea Dmean (left, -6.05 ± 4.33 Gy; right, -4.84 ± 4.66 Gy), lower left and right parotid V20Gy (left, -6.45% ± 5.32%; right, -6.92% ± 3.45%), and a lower integral dose (-0.19 ± 0.19 Gy). However, RPP plans increased the Dmax in the body outside of CTV (body-CTV) (1.2 ± 1.43 Gy), indicating a slightly higher hotspot produced by the RPP plans.

CONCLUSION

IMPT plans generated by a broad-scope RPP model have a quality that is, at minimum, comparable with, and at times superior to, that of the expert plans. The RPP plans demonstrated a greater robustness for CTV coverage and better sparing for several OARs.

Machine learning for dose-volume histogram based clinical decision-making support system in radiation therapy plans for brain tumors

•

Extraction, analysis, and interpretation of historical treatment planning data is valuable but very time-consuming.

•

Proposed machine learning model classifies radiotherapy plans based on their treatment planning objectives and trade-offs.

•

Application of double nested cross-validation enabled to build a robust model that achieved 94% accuracy on a testing data.

•

Model reasoning investigated with SHAP values showed consistency with clinical observations.

Purpose

To create and investigate a novel, clinical decision-support system using machine learning (ML).

Methods and Materials

The ML model was developed based on 79 radiotherapy plans of brain tumor patients that were prescribed a total dose of 60 Gy delivered with volumetric-modulated arc therapy (VMAT). Structures considered for analysis included planning target volume (PTV), brainstem, cochleae, and optic chiasm. The model aimed to classify the target variable that included class-0 corresponding to plans for which the PTV treatment planning objective was met and class-1 that was associated with plans for which the PTV objective was not met due to the priority trade-off to meet one or more organs-at-risk constraints. Several models were evaluated using double-nested cross-validation and an area-under-the-curve (AUC) metric, with the highest performing one selected for further investigation. The model predictions were explained with Shapely additive explanation (SHAP) interaction values.

Results

The highest-performing model was Logistic Regression achieving an accuracy of 93.8 ± 4.1% and AUC of 0.98 ± 0.02 on the testing data. The SHAP analysis indicated that the ΔD_99% metric for PTV had the greatest influence on the model predictions. The least important feature was ΔD_MAX for the left and right cochleae.

Conclusions

The trained model achieved satisfactory accuracy and can be used by medical physicists in a data-driven quality assurance program as well as by radiation oncologists to support their decision-making process in terms of treatment plan approval and potential plan modifications. Model explanation analysis showed that the model relies on clinically valid logic when making predictions.

Fast generation of lung SBRT plans with a knowledge-based planning model on ring-mounted Halcyon Linac

Purpose

To demonstrate fast treatment planning feasibility of stereotactic body radiation therapy (SBRT) for centrally located lung tumors on Halcyon Linac via a previously validated knowledge‐based planning (KBP) model to support offline adaptive radiotherapy.

Materials/methods

Twenty previously treated non‐coplanar volumetric‐modulated arc therapy (VMAT) lung SBRT plans (c‐Truebeam) on SBRT‐dedicated C‐arm Truebeam Linac were selected. Patients received 50 Gy in five fractions. c‐Truebeam plans were re‐optimized for Halcyon manually (m‐Halcyon) and with KBP model (k‐Halcyon). Both m‐Halcyon and k‐Halcyon plans were normalized for identical or better target coverage than clinical c‐Truebeam plans and compared for target conformity, dose heterogeneity, dose fall‐off, and dose tolerances to the organs‐at‐risk (OAR). Treatment delivery parameters and planning times were evaluated.

Results

k‐Halcyon plans were dosimetrically similar or better than m‐Halcyon and c‐Truebeam plans. k‐Halcyon and m‐Halcyon plan comparisons are presented with respect to c‐Truebeam. Differences in conformity index were statistically insignificant in k‐Halcyon and on average 0.02 higher (p = 0.04) in m‐Halcyon plans. Gradient index was on average 0.43 (p = 0.006) lower and 0.27 (p = 0.02) higher for k‐Halcyon and m‐Halcyon, respectively. Maximal dose 2 cm away in any direction from target was statistically insignificant. k‐Halcyon increased maximal target dose on average by 2.9 Gy (p < 0.001). Mean lung dose was on average reduced by 0.10 Gy (p = 0.004) in k‐Halcyon and increased by 0.14 Gy (p < 0.001) in m‐Halcyon plans. k‐Halcyon plans lowered bronchial tree dose on average by 1.2 Gy. Beam‐on‐time (BOT) was increased by 2.85 and 1.67 min, on average for k‐Halcyon and m‐Halcyon, respectively. k‐Halcyon plans were generated in under 30 min compared to estimated dedicated 180 ± 30 min for m‐Halcyon or c‐Truebeam plan.

Conclusion

k‐Halcyon plans were generated in under 30 min with excellent plan quality. This adaptable KBP model supports high‐volume clinics in the expansion or transfer of lung SBRT patients to Halcyon.

Knowledge-Based Planning for Intact Cervical Cancer

Cervical cancer radiotherapy is often complicated by significant variability in the quality and consistency of treatment plans. Knowledge-based planning (KBP), which utilizes prior patient data to correlated achievable optimal dosimetry with patient-specific anatomy, has demonstrated promise as a quality control tool for controlling this variability, with consequences for patient outcomes, as well as for the reliability of data from multi-institutional clinical trials. In this article we highlight the application of KBP-based quality control to cervical cancer radiotherapy. We discuss the potential impact of KBP on multi-institutional clinical trials to standardize cervical cancer treatment planning across diverse clinics, and discuss challenges and progress in the implementation of KBP for brachytherapy treatment planning. Additionally, we briefly discuss secondary applications of KBP for cervical cancer. The emerging picture from these studies indicates several exciting opportunities for increasing the utilization of KBP in day-to-day cervical cancer radiotherapy.

Personalising treatment plan quality review with knowledge-based planning in the TROG 15.03 trial for stereotactic ablative body radiotherapy in primary kidney cancer

INTRODUCTION

Quality assurance (QA) of treatment plans in clinical trials improves protocol compliance and patient outcomes. Retrospective use of knowledge-based-planning (KBP) in clinical trials has demonstrated improved treatment plan quality and consistency. We report the results of prospective use of KBP for real-time QA of treatment plan quality in the TROG 15.03 FASTRACK II trial, which evaluates efficacy of stereotactic ablative body radiotherapy (SABR) for kidney cancer.

METHODS

A KBP model was generated based on single institution data. For each patient in the KBP phase (open to the last 31 patients in the trial), the treating centre submitted treatment plans 7 days prior to treatment. A treatment plan was created by using the KBP model, which was compared with the submitted plan for each organ-at-risk (OAR) dose constraint. A report comparing each plan for each OAR constraint was provided to the submitting centre within 24 h of receiving the plan. The centre could then modify the plan based on the KBP report, or continue with the existing plan.

RESULTS

Real-time feedback using KBP was provided in 24/31 cases. Consistent plan quality was in general achieved between KBP and the submitted plan. KBP review resulted in replan and improvement of OAR dosimetry in two patients. All centres indicated that the feedback was a useful QA check of their treatment plan.

CONCLUSION

KBP for real-time treatment plan review was feasible for 24/31 cases, and demonstrated ability to improve treatment plan quality in two cases. Challenges include integration of KBP feedback into clinical timelines, interpretation of KBP results with respect to clinical trade-offs, and determination of appropriate plan quality improvement criteria.

Knowledge-based radiation treatment planning: A data-driven method survey

This paper surveys the data-driven dose prediction methods investigated for knowledge-based planning (KBP) in the last decade. These methods were classified into two major categories-traditional KBP methods and deep-learning (DL) methods-according to their techniques of utilizing previous knowledge. Traditional KBP methods include studies that require geometric or anatomical features to either find the best-matched case(s) from a repository of prior treatment plans or to build dose prediction models. DL methods include studies that train neural networks to make dose predictions. A comprehensive review of each category is presented, highlighting key features, methods, and their advancements over the years. We separated the cited works according to the framework and cancer site in each category. Finally, we briefly discuss the performance of both traditional KBP methods and DL methods, then discuss future trends of both data-driven KBP methods to dose prediction.

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.

Machine learning applications in radiation oncology

Machine learning technology has a growing impact on radiation oncology with an increasing presence in research and industry. The prevalence of diverse data including 3D imaging and the 3D radiation dose delivery presents potential for future automation and scope for treatment improvements for cancer patients. Harnessing this potential requires standardization of tools and data, and focused collaboration between fields of expertise. The rapid advancement of radiation oncology treatment technologies presents opportunities for machine learning integration with investments targeted towards data quality, data extraction, software, and engagement with clinical expertise. In this review, we provide an overview of machine learning concepts before reviewing advances in applying machine learning to radiation oncology and integrating these techniques into the radiation oncology workflows. Several key areas are outlined in the radiation oncology workflow where machine learning has been applied and where it can have a significant impact in terms of efficiency, consistency in treatment and overall treatment outcomes. This review highlights that machine learning has key early applications in radiation oncology due to the repetitive nature of many tasks that also currently have human review. Standardized data management of routinely collected imaging and radiation dose data are also highlighted as enabling engagement in research utilizing machine learning and the ability integrate these technologies into clinical workflow to benefit patients. Physicists need to be part of the conversation to facilitate this technical integration.

Development and clinical validation of Knowledge-based planning for Volumetric Modulated Arc Therapy of cervical cancer including pelvic and para aortic fields

•

A knowledge-based planning model was configured for VMAT of cervical cancer.

•

Knowledge-based plans were comparable, and for some OARs, outperformed clinical plans.

•

Improved organ sparing was observed, when individual patient geometry was considered.

Background and Purpose

Knowledge-based planning (KBP) is based on a model to estimate dose-volume histograms, configured using a library of historical treatment plans to efficiently create high quality plans. The aim was to report configuration and validation of KBP for Volumetric Modulated Arc Therapy of cervical cancer.

Materials and methods

A KBP model was configured from the institutional database (n = 125), including lymph node positive (n = 60) and negative (n = 65) patients. KBP Predicted plans were compared with Clinical Plans (CP) and Re-plans (Predicted plan as a base-plan) to validate the model. Model quality was quantified using coefficient of determination R2, mean square error (MSE), standard two-tailed paired t-test and Wilcoxon signed rank test.

Results

Estimation capability of the model was good for the bowel bag (MSE = 0.001, R2 = 0.84), modest for the bladder (MSE = 0.008) and poor for the rectum (MSE = 0.02 R2 = 0.78). KBP resulted in comparable target coverage, superior organ sparing as compared to CP. Re-plans outperformed CP for the bladder, V30 (66 ± 11% vs 74 ± 11%, p < .001), V40 (48 ± 14% vs 52 ± 14%, p < .001), however sparing was modest for the bowel bag V30 (413 ± 191cm³ vs 445 ± 208cm³, p = .037) V40 (199 ± 105cm³ vs 218 ± 127cm³, p = .031). All plans were comparable for rectum, while KBP resulted in significant sparing for spinal cord, kidneys and femoral heads.

Conclusion

KBP yielded comparable and for some organs superior performance compared to CP resulting in conformal and homogeneous target coverage. Improved organ sparing was observed when individual patient geometry was considered.

Prospective study of artificial intelligence-based decision support to improve head and neck radiotherapy plan quality

•

H&N radiation treatment plan directives are typically not patient-specific.

•

Patient-specific directives may facilitate the best-achievable dose distribution.

•

Use of an AI-guided tool significantly improved achieved dose for nearly all OARs.

Background and purpose

Volumetric modulated arc therapy (VMAT) planning for head and neck cancer is a complex process. While the lowest achievable dose for each individual organ-at-risk (OAR) is unknown a priori, artificial intelligence (AI) holds promise as a tool to accurately estimate the expected dose distribution for OARs. We prospectively investigated the benefits of incorporating an AI-based decision support tool (DST) into the clinical workflow to improve OAR sparing.

Materials and methods

The DST dose prediction model was based on 276 institutional VMAT plans. Under an IRB-approved prospective trial, the physician first generated a custom OAR directive for 50 consecutive patients (physician directive, PD). The DST then estimated OAR doses (AI directive, AD). For each OAR, the treating physician used the lower directive to form a hybrid directive (HD). The final plan metrics were compared to each directive. A dose difference of 3 Gray (Gy) was considered clinically significant.

Results

Compared to the AD and PD, the HD reduced OAR dose objectives by more than 3 Gy in 22% to 75% of cases, depending on OAR. The resulting clinical plan typically met these lower constraints and achieved mean dose reductions between 4.3 and 16 Gy over the PD, and 5.6 to 9.1 Gy over the AD alone. Dose metrics achieved using the HD were significantly better than institutional historical plans for most OARs and NRG constraints for all OARs.

Conclusions

The DST facilitated a significantly improved treatment directive across all OARs for this generalized H&N patient cohort, with neither the AD nor PD alone sufficient to optimally direct planning.

Evaluation of a highly refined prediction model in knowledge-based volumetric modulated arc therapy planning for cervical cancer

Background and purpose

To explore whether a highly refined dose volume histograms (DVH) prediction model can improve the accuracy and reliability of knowledge-based volumetric modulated arc therapy (VMAT) planning for cervical cancer.

Methods and materials

The proposed model underwent repeated refining through progressive training until the training samples increased from initial 25 prior plans up to 100 cases. The estimated DVHs derived from the prediction models of different runs of training were compared in 35 new cervical cancer patients to analyze the effect of such an interactive plan and model evolution method. The reliability and efficiency of knowledge-based planning (KBP) using this highly refined model in improving the consistency and quality of the VMAT plans were also evaluated.

Results

The prediction ability was reinforced with the increased number of refinements in terms of normal tissue sparing. With enhanced prediction accuracy, more than 60% of automatic plan-6 (AP-6) plans (22/35) can be directly approved for clinical treatment without any manual revision. The plan quality scores for clinically approved plans (CPs) and manual plans (MPs) were on average 89.02 ± 4.83 and 86.48 ± 3.92 (p < 0.001). Knowledge-based planning significantly reduced the D_mean and V_{18 Gy} for kidney (L/R), the D_mean, V_{30 Gy}, and V_{40 Gy} for bladder, rectum, and femoral head (L/R).

Conclusion

The proposed model evolution method provides a practical way for the KBP to enhance its prediction ability with minimal human intervene. This highly refined prediction model can better guide KBP in improving the consistency and quality of the VMAT plans.

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.