Automation and artificial intelligence in radiation therapy treatment planning

Automation and artificial intelligence (AI) is already possible for many radiation therapy planning and treatment processes with the aim of improving workflows and increasing efficiency in radiation oncology departments. Currently, AI technology is advancing at an exponential rate, as are its applications in radiation oncology. This commentary highlights the way AI has begun to impact radiation therapy treatment planning and looks ahead to potential future developments in this space. Historically, radiation therapist's (RT's) role has evolved alongside the adoption of new technology. In Australia, RTs have key clinical roles in both planning and treatment delivery and have been integral in the implementation of automated solutions for both areas. They will need to continue to be informed, to adapt and to transform with AI technologies implemented into clinical practice in radiation oncology departments. RTs will play an important role in how AI-based automation is implemented into practice in Australia, ensuring its application can truly enable personalised and higher-quality treatment for patients. To inform and optimise utilisation of AI, research should not only focus on clinical outcomes but also AI's impact on professional roles, responsibilities and service delivery. Increased efficiencies in the radiation therapy workflow and workforce need to maintain safe improvements in practice and should not come at the cost of creativity, innovation, oversight and safety.

Development of automated region of interest selection algorithms for surface-guided radiation therapy of breast cancer

To investigate automation of the preparation of the region of interest (ROI) for surface-guided radiotherapy (SGRT) of the whole breast with two algorithms based on contour anatomies: using the body contour, and using the breast contour. The patient dataset used for modeling consisted of 39 breast cancer patients previously treated with SGRT. The patient's anatomical structures (body and ipsilateral breast) were retrieved from the planning system, and the clinical ROI (cROI) drawn by the planners was retrieved from the SGRT system for comparison. For the body-contour-based algorithm, a convolutional neural network (MobileNet-v2) was utilized to train a synthetic human model dataset to predict body joint locations. With the body joint location knowledge, an automated ROI (aROIbody ) can be created based on: (1) the superior-inferior (S-I) borders defined by the joint locations, (2) the left-right (L-R) borders defined with 3/4 of chest width, and (3) a curation of the ROI to avoid the ipsilateral armpit. For the breast-contour-based algorithm, an aROIbreast was created by first defining the ROI in the S-I direction with the ipsilateral breast boundaries. Other steps are the same as with the body-contour-based algorithm. Among the 39 patients, 24 patients were used to fine-tune the algorithm parameters, and the remaining 15 patients were used to evaluate the quality of the aROIs against the cROIs. A blinded evaluation was performed by three SGRT expert physicists to rate the acceptability and the quality (1-10 scale) of the aROIs and cROIs, and the dice similarity coefficient (DSC) was also calculated to compare the similarity between the aROIs and cROIs. The results showed that the average acceptability was 14/15 (range: 13/15-15/15) for cROIs, 13.3/15 (range: 13/15-14/15) for aROIbody , and 14.6/15 (range: 14/15-15/15) for aROIbreast . The average quality was 7.4 ± 0.8 for cROIs, 8.1 ± 1.2 for aROIbody , and 8.2 ± 0.9 for aROIbreast . The DSC with cROIs was 0.81 ± 0.06 for aROIbody , and 0.83 ± 0.04 for aROIbreast . The ROI creation time was ∼120 s for clinical, 1.3 s for aROIbody , and 1.2 s for aROIbreast . The proposed automated algorithms can improve the ROI compliance with the SGRT protocol, with a shortened preparation time. It is ready to be integrated into the clinical workflow for automated ROI preparation.

Automatic quality assurance of radiotherapy treatment plans using Bayesian networks: A multi-institutional study

Purpose

Artificial intelligence applications in radiation oncology have been the focus of study in the last decade. The introduction of automated and intelligent solutions for routine clinical tasks, such as treatment planning and quality assurance, has the potential to increase safety and efficiency of radiotherapy. In this work, we present a multi-institutional study across three different institutions internationally on a Bayesian network (BN)-based initial plan review assistive tool that alerts radiotherapy professionals for potential erroneous or suboptimal treatment plans.

Methods

Clinical data were collected from the oncology information systems in three institutes in Europe (Maastro clinic - 8753 patients treated between 2012 and 2020) and the United States of America (University of Vermont Medical Center [UVMMC] - 2733 patients, University of Washington [UW] - 6180 patients, treated between 2018 and 2021). We trained the BN model to detect potential errors in radiotherapy treatment plans using different combinations of institutional data and performed single-site and cross-site validation with simulated plans with embedded errors. The simulated errors consisted of three different categories: i) patient setup, ii) treatment planning and iii) prescription. We also compared the strategy of using only diagnostic parameters or all variables as evidence for the BN. We evaluated the model performance utilizing the area under the receiver-operating characteristic curve (AUC).

Results

The best network performance was observed when the BN model is trained and validated using the dataset in the same center. In particular, the testing and validation using UVMMC data has achieved an AUC of 0.92 with all parameters used as evidence. In cross-validation studies, we observed that the BN model performed better when it was trained and validated in institutes with similar technology and treatment protocols (for instance, when testing on UVMMC data, the model trained on UW data achieved an AUC of 0.84, compared with an AUC of 0.64 for the model trained on Maastro data). Also, combining training data from larger clinics (UW and Maastro clinic) and using it on smaller clinics (UVMMC) leads to satisfactory performance with an AUC of 0.85. Lastly, we found that in general the BN model performed better when all variables are considered as evidence.

Conclusion

We have developed and validated a Bayesian network model to assist initial treatment plan review using multi-institutional data with different technology and clinical practices. The model has shown good performance even when trained on data from clinics with divergent profiles, suggesting that the model is able to adapt to different data distributions.

Automated testing platform for radiotherapy treatment planning scripts

Realizing the potential of user-developed automation software interacting with a treatment planning system (TPS) requires rigorous testing to ensure patient safety and data integrity. We developed an automated test platform to allow comparison of the treatment planning database before and after the execution of a write-enabled script interacting with a commercial TPS (Eclipse, Varian Medical Systems, Palo Alto, CA) using the vendor-provided Eclipse Scripting Application Programming Interface (ESAPI). The C#-application known as Write-Enable Script Testing Engine (WESTE) serializes the treatment planning objects (Patient, Structure Set, PlanSetup) accessible through ESAPI, and then compares the serialization acquired before and after the execution of the script being tested, documenting identified differences to highlight the changes made to the treatment planning data. The first two uses of WESTE demonstrated that the testing platform could acquire and analyze the data quickly (<4 s per test case) and facilitate the clinical implementation of write-enabled scripts.

Using feasibility dose-volume histograms to reduce intercampus plan quality variability for head-and-neck cancer

The purpose of this work is to objectively assess variability of intercampus plan quality for head-and-neck (HN) cancer and to test utility of a priori feasibility dose-volume histograms (FDVHs) as planning dose goals. In this study, 109 plans treated from 2017 to 2019 were selected, with 52 from the main campus and 57 from various regional centers. For each patient, the planning computed tomography images and contours were imported into a commercial program to generate FDVHs with a feasibility value (f-value) ranging from 0.0 to 0.5. For 10 selected organs-at-risk (OARs), we used the Dice similarity coefficient (DSC) to quantify the overlaps between FDVH and clinically achieved DVH of each OAR and determined the f-value associated with the maximum DSC (labeled as f-max). Subsequently, 10 HN plans from the regional centers were replanned with planning dose goals guided by FDVHs. The clinical and feasibility-guided auto-planning (FgAP) plans were evaluated using our institutional criteria. Among plans from the main campus and regional centers, the median f-max values were statistically significantly different (p < 0.05) for all OARs except for the left parotid (p = 0.622), oral cavity (p = 0.057), and mandible (p = 0.237). For the 10 FgAP plans, the median values of f-max were 0.21, compared to 0.37 from the clinical plans. With comparable dose coverage to the tumor volumes, the significant differences (p < 0.05) in the median f-max and corresponding dose reduction (shown in parenthesis) for the spinal cord, larynx, supraglottis, trachea, and esophagus were 0.27 (8.5 Gy), 0.3 (7.6 Gy), 0.19 (5.9 Gy), 0.19 (8.9 Gy), and 0.12 (4.0 Gy), respectively. In conclusion, the FDVH prediction is an objective quality assurance tool to evaluate the intercampus plan variability. This tool can also provide guideline in planning dose goals to further improve plan quality.

MR-LINAC-Guided Adaptive Radiotherapy for Gastric MALT: Two Case Reports and a Literature Review

It is still very challenging to use conventional radiation therapy techniques to treat stomach tumors, although image-guided radiotherapy, mainly by kV X-ray imaging techniques, has become routine in the clinic. This is because the stomach is one of the most deformable organs, and thus it is vulnerable to respiratory motions, daily diet, and body position changes. In addition, X-ray radiographs and CT volumetric images have low contrast in soft tissues. In contrast, magnetic resonance imaging (MRI) techniques provide good contrast in images of soft tissues. The emerging MR-guided radiotherapy, based on the MR-LINAC system, may have the potential to solve the above difficulties due to its unique advantages. The real-time imaging feature and the high-contrast of soft tissues MR images provided by the MR-LINAC system have facilitated the therapeutic adaptive planning. Online learning capabilities could be used to optimize the automatic delineation of the target organ or tissue prior to each radiotherapy session. This could greatly improve the accuracy and efficiency of the target delineation in adaptive planning. In this clinical case report, we elaborated a workflow for the diagnosis and treatment of two patients with gastric mucosa-associated lymphoid tissue (MALT) lymphoma. One patient underwent MR-guided daily adaptive radiotherapy based on daily automated segmentation using the novel artificial intelligence (AI) technique for gastric delineation.

Barriers and facilitators to clinical implementation of radiotherapy treatment planning automation: A survey study of medical dosimetrists

PURPOSE

Little is known about the scale of clinical implementation of automated treatment planning techniques in the United States. In this work, we examine the barriers and facilitators to adoption of commercially available automated planning tools into the clinical workflow using a survey of medical dosimetrists.

METHODS/MATERIALS

Survey questions were developed based on a literature review of automation research and cognitive interviews of medical dosimetrists at our institution. Treatment planning automation was defined to include auto-contouring and automated treatment planning. Survey questions probed frequency of use, positive and negative perceptions, potential implementation changes, and demographic and institutional descriptive statistics. The survey sample was identified using both a LinkedIn search and referral requests sent to physics directors and senior physicists at 34 radiotherapy clinics in our state. The survey was active from August 2020 to April 2021.

RESULTS

Thirty-four responses were collected out of 59 surveys sent. Three categories of barriers to use of automation were identified. The first related to perceptions of limited accuracy and usability of the algorithms. Eighty-eight percent of respondents reported that auto-contouring inaccuracy limited its use, and 62% thought it was difficult to modify an automated plan, thus limiting its usefulness. The second barrier relates to the perception that automation increases the probability of an error reaching the patient. Third, respondents were concerned that automation will make their jobs less satisfying and less secure. Large majorities reported that they enjoyed plan optimization, would not want to lose that part of their job, and expressed explicit job security fears.

CONCLUSION

To our knowledge this is the first systematic investigation into the views of automation by medical dosimetrists. Potential barriers and facilitators to use were explicitly identified. This investigation highlights several concrete approaches that could potentially increase the translation of automation into the clinic, along with areas of needed research.

The Challenges Facing Deep Learning based Catheter Localization for Ultrasound Guided High-Dose-Rate Prostate Brachytherapy

BACKGROUND

Automated catheter localization for ultrasound guided high-dose-rate prostate brachytherapy faces challenges relating to imaging noise and artifacts. To date, catheter reconstruction during the clinical procedure is performed manually. Deep learning has been successfully applied to a wide variety of complex tasks and has the potential to tackle the unique challenges associated with multiple catheter localization on ultrasound. Such a task is well suited for automation, with the potential to improve productivity and reliability.

PURPOSE

We developed a deep learning model for automated catheter reconstruction and investigated potential factors influencing model performance. The model was designed to integrate into a clinical workflow, with a proposed reconstruction confidence metric to aid in planner verification.

METHODS

Datasets from 242 patients treated from 2016 to 2020 were collected retrospectively. The anonymized dataset comprises of 31,000 transverse images reconstructed from 3D sagittal ultrasound acquisitions and 3,500 implanted catheters manually localized by the planner. Each catheter was retrospectively ranked based on the severity of imaging artifacts affecting reconstruction difficulty. The U-NET deep learning architecture was trained to localize implanted catheters on transverse images. A five-fold cross-validation method was used, allowing for evaluation over the entire dataset. The post-processing software combined the predictions with patient-specific implant information to reconstructed catheters in 3D space, uniquely matched to the implanted grid positions. A reconstruction confidence metric was calculated based on the number and probability of localized predictions per catheter. For each patient, deep learning prediction and post-processing reconstruction was completed in under two minutes on a non-performance PC.

RESULTS

Overall, 80% of catheter reconstructions were accurate, within 2 mm along 90% of the length. The catheter tip was often not detected and required extrapolation during reconstruction. The reconstruction accuracy was 89% for the easiest catheter ranking and decreased to 13% for the highest difficulty ranking, when the aid of live ultrasound would have been recommended. Even when limited to the easiest ranked catheters, the reconstruction accuracy decreased at distal grid positions, down to 50%. Individual implantation style was found to influence the frequency of severe artifacts, slightly impacting the model accuracy. A reconstruction confidence metric identified the difficult catheters, removed the observed individual variation, and increased the overall accuracy to 91% while excluding 27% of the reconstructions.

CONCLUSIONS

The deep learning model localized implanted catheters over a large clinical dataset, with overall promising results. The model faced challenges due to ultrasound artifacts and image degradation distal to the probe, underlining the continued importance of maintaining image quality and minimizing artifacts. A potential workflow for integration into the clinical procedure was demonstrated, including the use of a confidence metric to predict low accuracy reconstructions. Comparison between models evaluated on different datasets should also consider underlying differences, such as the frequency and severity of imaging artifacts. This article is protected by copyright. All rights reserved.

Improving the Quality of Care in Radiation Oncology using Artificial Intelligence

Radiation therapy is a complex process involving multiple professionals and steps from simulation to treatment planning to delivery, and these procedures are prone to error. Additionally, the imaging and treatment delivery equipment in radiotherapy is highly complex and interconnected and represents another risk point in the quality of care. Numerous quality assurance tasks are carried out to ensure quality and to detect and prevent potential errors in the process of care. Recent developments in artificial intelligence provide potential tools to the radiation oncology community to improve the efficiency and performance of quality assurance efforts. Targets for artificial intelligence enhancement include the quality assurance of treatment plans, target and tissue structure delineation used in the plans, delivery of the plans and the radiotherapy delivery equipment itself. Here we review recent developments of artificial intelligence applications that aim to improve quality assurance processes in radiation therapy and discuss some of the challenges and limitations that require further development work to realise the potential of artificial intelligence for quality assurance.

ORBIT-RT: A Real-Time, Open Platform for Knowledge-Based Quality Control of Radiotherapy Treatment Planning

PURPOSE

Access to knowledge-based treatment plan quality control has been hindered by the complexity of developing models and integration with different treatment planning systems (TPS). Online Real-time Benchmarking Information Technology for RadioTherapy (ORBIT-RT) provides a free, web-based platform for knowledge-based dose estimation that can be used by clinicians worldwide to benchmark the quality of their radiotherapy plans.

MATERIALS AND METHODS

The ORBIT-RT platform was developed to satisfy four primary design criteria: web-based access, TPS independence, Health Insurance Portability and Accountability Act compliance, and autonomous operation. ORBIT-RT uses a cloud-based server to automatically anonymize a user's Digital Imaging and Communications in Medicine for RadioTherapy (DICOM-RT) file before upload and processing of the case. From there, ORBIT-RT uses established knowledge-based dose-volume histogram (DVH) estimation methods to autonomously create DVH estimations for the uploaded DICOM-RT. ORBIT-RT performance was evaluated with an independent validation set of 45 volumetric modulated arc therapy prostate plans with two key metrics: (i) accuracy of the DVH estimations, as quantified by their error, DVH_clinical - DVH_prediction and (ii) time to process and display the DVH estimations on the ORBIT-RT platform.

RESULTS

ORBIT-RT organ DVH predictions show < 1% bias and 3% error uncertainty at doses > 80% of prescription for the prostate validation set. The ORBIT-RT extensions require 3.0 seconds per organ to analyze. The DICOM upload, data transfer, and DVH output display extend the entire system workflow to 2.5-3 minutes.

CONCLUSION

ORBIT-RT demonstrated fast and fully autonomous knowledge-based feedback on a web-based platform that takes only anonymized DICOM-RT as input. The ORBIT-RT system can be used for real-time quality control feedback that provides users with objective comparisons for final plan DVHs.

Personalized Treatment Planning Automation in Prostate Cancer Radiation Oncology: A Comprehensive Dosimetric Study

Background

In radiation oncology, automation of treatment planning has reported the potential to improve plan quality and increase planning efficiency. We performed a comprehensive dosimetric evaluation of the new Personalized algorithm implemented in Pinnacle³ for full planning automation of VMAT prostate cancer treatments.

Material and Methods

Thirteen low-risk prostate (without lymph-nodes irradiation) and 13 high-risk prostate (with lymph-nodes irradiation) treatments were retrospectively taken from our clinical database and re-optimized using two different automated engines implemented in the Pinnacle treatment system. These two automated engines, the currently used Autoplanning and the new Personalized are both template-based algorithms that use a wish-list to formulate the planning goals and an iterative approach able to mimic the planning procedure usually adopted by experienced planners. In addition, the new Personalized module integrates a new engine, the Feasibility module, able to generate an “a priori” DVH prediction of the achievability of planning goals. Comparison between clinically accepted manually generated (MP) and automated plans generated with both Autoplanning (AP) and Personalized engines (Pers) were performed using dose-volume histogram metrics and conformity indexes. Three different normal tissue complication probabilities (NTCPs) models were used for rectal toxicity evaluation. The planning efficiency and the accuracy of dose delivery were assessed for all plans.

Results

For similar targets coverage, Pers plans reported a significant increase of dose conformity and less irradiation of healthy tissue, with significant dose reduction for rectum, bladder, and femurs. On average, Pers plans decreased rectal mean dose by 11.3 and 8.3 Gy for low-risk and high-risk cohorts, respectively. Similarly, the Pers plans decreased the bladder mean doses by 7.3 and 7.6 Gy for low-risk and high-risk cohorts, respectively. The integral dose was reduced by 11–16% with respect to MP plans. Overall planning times were dramatically reduced to about 7 and 15 min for Pers plans. Despite the increased complexity, all plans passed the 3%/2 mm γ-analysis for dose verification.

Conclusions

The Personalized engine provided an overall increase of plan quality, in terms of dose conformity and sparing of normal tissues for prostate cancer patients. The Feasibility “a priori” DVH prediction module provided OARs dose sparing well beyond the clinical objectives. The new Pinnacle Personalized algorithms outperformed the currently used Autoplanning ones as solution for treatment planning automation.

A practical method to quantify knowledge-based DVH prediction accuracy and uncertainty with reference cohorts

The adoption of knowledge-based dose-volume histogram (DVH) prediction models for assessing organ-at-risk (OAR) sparing in radiotherapy necessitates quantification of prediction accuracy and uncertainty. Moreover, DVH prediction error bands should be readily interpretable as confidence intervals in which to find a percentage of clinically acceptable DVHs. In the event such DVH error bands are not available, we present an independent error quantification methodology using a local reference cohort of high-quality treatment plans, and apply it to two DVH prediction models, ORBIT-RT and RapidPlan, trained on the same set of 90 volumetric modulated arc therapy (VMAT) plans. Organ-at-risk DVH predictions from each model were then generated for a separate set of 45 prostate VMAT plans. Dose-volume histogram predictions were then compared to their analogous clinical DVHs to define prediction errorsV c l i n , i - V p r e d , i (ith plan), from which prediction bias μ, prediction error variation σ, and root-mean-square error R M S E pred ≡ 1 N ∑ i V c l i n , i - V p r e d , i 2 ≅ σ 2 + μ 2 could be calculated for the cohort. The empirical R M S E pred was then contrasted to the model-provided DVH error estimates. For all prostate OARs, above 50% Rx dose, ORBIT-RT μ and σ were comparable to or less than those of RapidPlan. Above 80% Rx dose, μ < 1% and σ < 3-4% for both models. As a result, above 50% Rx dose, ORBIT-RT R M S E pred was below that of RapidPlan, indicating slightly improved accuracy in this cohort. Because μ ≈ 0, R M S E pred is readily interpretable as a canonical standard deviation σ, whose error band is expected to correctly predict 68% of normally distributed clinical DVHs. By contrast, RapidPlan's provided error band, although described in literature as a standard deviation range, was slightly less predictive than R M S E pred (55-70% success), while the provided ORBIT-RT error band was confirmed to resemble an interquartile range (40-65% success) as described. Clinicians can apply this methodology using their own institutions' reference cohorts to (a) independently assess a knowledge-based model's predictive accuracy of local treatment plans, and (b) interpret from any error band whether further OAR dose sparing is likely attainable.

Integration of automation into an existing clinical workflow to improve efficiency and reduce errors in the manual treatment planning process for total body irradiation (TBI)

Purpose

To identify causes of error, and present the concept of an automated technique that improves efficiency and helps to reduce transcription and manual data entry errors in the treatment planning of total body irradiation (TBI).

Methods

Analysis of incidents submitted to incident learning system (ILS) was performed to identify potential avenues for improvement by implementation of automation of the manual treatment planning process for total body irradiation (TBI). Following this analysis, it became obvious that while the individual components of the TBI treatment planning process were well implemented, the manual ‘bridging’ of the components (transcribing data, manual data entry etc.) were leading to high potential for error. A C#‐based plug‐in treatment planning script was developed to remove the manual parts of the treatment planning workflow that were contributing to increased risk.

Results

Here we present an example of the implementation of “Glue” programming, combining treatment planning C# scripts with existing spreadsheet calculation worksheets. Prior to the implementation of automation, 35 incident reports related to the TBI treatment process were submitted to the ILS over a 6‐year period, with an average of 1.4 ± 1.7 reports submitted per quarter. While no incidents reached patients, reports ranged from minor documentation issues to potential for mistreatment if not caught before delivery. Since the implementation of automated treatment planning and documentation, treatment planning time per patient, including documentation, has been reduced; from an average of 45 min pre‐automation to <20 min post‐automation.

Conclusions

Manual treatment planning techniques may be well validated, but they are time‐intensive and have potential for error. Often the barrier to automating these techniques becomes the time required to “re‐code” existing solutions in unfamiliar computer languages. We present the workflow here as a proof of concept that automation may help to improve clinical efficiency and safety for special procedures.

Automated treatment planning of postmastectomy radiotherapy

Purpose

Breast cancer is the most common cancer in women globally and radiation therapy is a cornerstone of its treatment. However, there is an enormous shortage of radiotherapy staff, especially in low‐ and middle‐income countries. This shortage could be ameliorated through increased automation in the radiation treatment planning process, which may reduce the workload on radiotherapy staff and improve efficiency in preparing radiotherapy treatments for patients. To this end, we sought to create an automated treatment planning tool for postmastectomy radiotherapy (PMRT).

Methods

Algorithms to automate every step of PMRT planning were developed and integrated into a commercial treatment planning system. The only required inputs for automated PMRT planning are a planning computed tomography scan, a plan directive, and selection of the inferior border of the tangential fields. With no other human input, the planning tool automatically creates a treatment plan and presents it for review. The major automated steps are (a) segmentation of relevant structures (targets, normal tissues, and other planning structures), (b) setup of the beams (tangential fields matched with a supraclavicular field), and (c) optimization of the dose distribution by using a mix of high‐ and low‐energy photon beams and field‐in‐field modulation for the tangential fields. This automated PMRT planning tool was tested with ten computed tomography scans of patients with breast cancer who had received irradiation of the left chest wall. These plans were assessed quantitatively using their dose distributions and were reviewed by two physicians who rated them on a three‐tiered scale: use as is, minor changes, or major changes. The accuracy of the automated segmentation of the heart and ipsilateral lung was also assessed. Finally, a plan quality verification tool was tested to alert the user to any possible deviations in the quality of the automatically created treatment plans.

Results

The automatically created PMRT plans met the acceptable dose objectives, including target coverage, maximum plan dose, and dose to organs at risk, for all but one patient for whom the heart objectives were exceeded. Physicians accepted 50% of the treatment plans as is and required only minor changes for the remaining 50%, which included the one patient whose plan had a high heart dose. Furthermore, the automatically segmented contours of the heart and ipsilateral lung agreed well with manually edited contours. Finally, the automated plan quality verification tool detected 92% of the changes requested by physicians in this review.

Conclusions

We developed a new tool for automatically planning PMRT for breast cancer, including irradiation of the chest wall and ipsilateral lymph nodes (supraclavicular and level III axillary). In this initial testing, we found that the plans created by this tool are clinically viable, and the tool can alert the user to possible deviations in plan quality. The next step is to subject this tool to prospective testing, in which automatically planned treatments will be compared with manually planned treatments.

Adapting automated treatment planning configurations across international centres for prostate radiotherapy

Background and purpose

Automated configurations are increasingly utilised for radiotherapy treatment planning. This study investigates whether automated treatment planning configurations are adaptable across clinics with different treatment planning protocols for prostate radiotherapy.

Material and methods

The study comprised three participating centres, each with pre-existing locally developed prostate AutoPlanning configurations using the Pinnacle³® treatment planning system. Using a three-patient training dataset circulated from each centre, centres modified local prostate configurations to generate protocol compliant treatment plans for the other two centres. Each centre applied modified configurations on validation datasets distributed from each centre (10 patients from 3 centres). Plan quality was assessed through DVH analysis and protocol compliance.

Results

All treatment plans were clinically acceptable, based off relevant treatment protocol. Automated planning configurations from Centre’s A and B recorded 2 and 18 constraint and high priority deviations respectively. Centre C configurations recorded no high priority deviations. Centre A configurations produced treatment plans with superior dose conformity across all patient PTVs (mean = 1.14) compared with Centre’s B and C (mean = 1.24 and 1.22). Dose homogeneity was consistent between all centre’s configurations (mean = 0.083, 0.077, and 0.083 respectively).

Conclusions

This study demonstrates that automated treatment planning configurations can be shared and implemented across multiple centres with simple adaptations to local protocols.

Developing Predictive or Prognostic Biomarkers for Charged Particle Radiotherapy

The response to radiotherapy can vary greatly among individuals, even though advances in technology allow for the highly localized placement of therapeutic doses of radiation to a tumor. This variability in patient response to radiation is biologically driven, but the individuality of tumor and healthy tissue biology are not used to create individual treatment plans. Biomarkers of radiosensitivity, whether intrinsic or from hypoxia, would move radiation oncology from precision medicine to precise, personalized medicine. Charged particle radiotherapy allows for even greater dose conformity, but the biological advantages of charged particle radiotherapy have not yet been cultivated. The development of biomarkers that would drive biologically based clinical trials, identify patients for whom charged particles are most appropriate, or aid in particle-selection strategies could be envisioned with appropriate biomarkers. Initially, biomarkers for low-linear energy transfer (LET) radiation responses should be tested against charged particles. Biomarkers of tumor radioresistance to low-LET radiations could be used to identify patients for whom the enhanced relative biological effectiveness (RBE) of charged particles would be more effective compared with low-LET radiations and those for whom specific DNA-repair inhibitors, in combination with charged particles, may also be appropriate. Furthermore, heavy charged particles can overcome the radioresistance of hypoxic tumors when used at the appropriate LET. Biomarkers for hypoxia could identify hypoxic tumors and, in combination with imaging, define hypoxic regions of a tumor for specific ion selection. Moreover, because of the enhanced RBE for charged particles, the risk for adverse healthy tissue effects may be greater, even though charged particles have greater tumor conformality. There are many validated healthy-tissue biomarkers available to test against charged particle exposures. Lastly, newer biological techniques, as well as newer bioinformatic and computational methods, are rapidly changing the landscape for biomarker identification, validation, and clinical trial design.

Auto- versus human-driven plan in mediastinal Hodgkin lymphoma radiation treatment

Background

Technological advances in Hodgkin lymphoma (HL) radiation therapy (RT) by high conformal treatments potentially increase control over organs-at-risk (OARs) dose distribution. However, plan optimization remains a time-consuming task with great operator dependent variability. Purpose of the present study was to devise a fully automated pipeline based on the Pinnacle³ Auto-Planning (AP) algorithm for treating female supradiaphragmatic HL (SHL) patients.

Methods

CT-scans of 10 female patients with SHL were considered. A “butterfly” (BF) volumetric modulated arc therapy was optimized using SmartArc module integrated in Pinnacle³ v. 9.10 using Collapsed Cone Convolution Superposition algorithm (30 Gy in 20 fractions). Human-driven (Manual-BF) and AP-BF optimization plans were generated. For AP, an optimization objective list of Planning Target Volume (PTV)/OAR clinical goals was first implemented, starting from a subset of 5 patients used for algorithm training. This list was then tested on the remaining 5 patients (validation set). In addition to the BF technique, the AP engine was applied to a 2 coplanar disjointed arc (AP-ARC) technique using the same objective list. For plan evaluation, dose-volume-histograms of PTVs and OARs were extracted; homogeneity and conformity indices (HI and CI), OARs dose-volume metrics and odds for different toxicity endpoints were computed. Non-parametric Friedman and Dunn tests were used to identify significant differences between groups.

Results

A single AP objective list for SHL was obtained. Compared to the manual plan, both AP-plans offer comparable CIs while AP-ARC also achieved comparable HIs. All plans fulfilled the clinical dose criteria set for OARs: both AP solutions performed at least as good as Manual-BF plan. In particular, AP-ARC outperformed AP-BF in terms of heart sparing involving a lower risk of coronary events and radiation-induced lung fibrosis. Hands-on planning time decreased by a factor of 10 using AP on average.

Conclusions

Despite the high interpatient PTV (size and position) variability, it was possible to set a standard SHL AP optimization list with a high level of generalizability. Using the implemented list, the AP module was able to limit OAR doses, producing clinically acceptable plans with stable quality without additional user input. Overall, the AP engine associated to the arc technique represents the best option for SHL.

Early detection of potential errors during patient treatment planning

Purpose

Data errors caught late in treatment planning require time to correct, resulting in delays up to 1 week. In this work, we identify causes of data errors in treatment planning and develop a software tool that detects them early in the planning workflow.

Methods

Two categories of errors were studied: data transfer errors and TPS errors. Using root cause analysis, the causes of these errors were determined. This information was incorporated into a software tool which uses ODBC‐SQL service to access TPS's Postgres and Mosaiq MSSQL databases for our clinic. The tool then uses a read‐only FTP service to scan the TPS unix file system for errors. Detected errors are reviewed by a physicist. Once confirmed, clinicians are notified to correct the error and educated to prevent errors in the future. Time‐cost analysis was performed to estimate the time savings of implementing this software clinically.

Results

The main errors identified were incorrect patient entry, missing image slice, and incorrect DICOM tag for data transfer errors and incorrect CT‐density table application, incorrect image as reference CT, and secondary image imported to incorrect patient for TPS errors. The software has been running automatically since 2015. In 2016, 84 errors were detected with the most frequent errors being incorrect patient entry (35), incorrect CT‐density table (17), and missing image slice (16). After clinical interventions to our planning workflow, the number of errors in 2017 decreased to 44. Time savings in 2016 with the software is estimated to be 795 h. This is attributed to catching errors early and eliminating the need to replan cases.

Conclusions

New QA software detects errors during planning, improving the accuracy and efficiency of the planning process. This important QA tool focused our efforts on the data communication processes in our planning workflow that need the most improvement.

Data Analysis in Radiotherapy Treatments

Radiotherapy is one of the main cancer treatments available today, together with chemotherapy and surgery. Radiotherapy treatments have to be planned for each patient in an individualized manner. The knowledge acquired from one single treatment can be used to improve the treatment planning and outcome of several other patients. In the last years, attention has been drawn to the added value of using data analysis for radiotherapy treatment planning, prediction of treatment outcomes, survival analysis and quality assurance. In this article, existing literature is reviewed.

Monte Carlo verification of radiotherapy treatments with CloudMC

Background

A new implementation has been made on CloudMC, a cloud-based platform presented in a previous work, in order to provide services for radiotherapy treatment verification by means of Monte Carlo in a fast, easy and economical way. A description of the architecture of the application and the new developments implemented is presented together with the results of the tests carried out to validate its performance.

Methods

CloudMC has been developed over Microsoft Azure cloud. It is based on a map/reduce implementation for Monte Carlo calculations distribution over a dynamic cluster of virtual machines in order to reduce calculation time. CloudMC has been updated with new methods to read and process the information related to radiotherapy treatment verification: CT image set, treatment plan, structures and dose distribution files in DICOM format. Some tests have been designed in order to determine, for the different tasks, the most suitable type of virtual machines from those available in Azure. Finally, the performance of Monte Carlo verification in CloudMC is studied through three real cases that involve different treatment techniques, linac models and Monte Carlo codes.

Results

Considering computational and economic factors, D1_v2 and G1 virtual machines were selected as the default type for the Worker Roles and the Reducer Role respectively. Calculation times up to 33 min and costs of 16 € were achieved for the verification cases presented when a statistical uncertainty below 2% (2σ) was required. The costs were reduced to 3–6 € when uncertainty requirements are relaxed to 4%.

Conclusions

Advantages like high computational power, scalability, easy access and pay-per-usage model, make Monte Carlo cloud-based solutions, like the one presented in this work, an important step forward to solve the long-lived problem of truly introducing the Monte Carlo algorithms in the daily routine of the radiotherapy planning process.

A method for a priori estimation of best feasible DVH for organs-at-risk: Validation for head and neck VMAT planning

PURPOSE

Despite improvements in optimization and automation algorithms, the quality of radiation treatment plans still varies dramatically. A tool that allows a priori estimation of the best possible sparing (Feasibility DVH, or FDVH) of an organ at risk (OAR) in high-energy photon planning may help reduce plan quality variability by deriving patient-specific OAR goals prior to optimization. Such a tool may be useful for (a) meaningfully evaluating patient-specific plan quality and (b) supplying best theoretically achievable DVH goals, thus pushing the solution toward automatic Pareto optimality. This work introduces such a tool and validates it for clinical Head and Neck (HN) datasets.

METHODS

To compute FDVH, first the targets are assigned uniform prescription doses, with no reference to any particular beam arrangement. A benchmark 3D dose built outside the targets is estimated using a series of energy-specific dose spread calculations reflecting observed properties of radiation distribution in media. For the patient, the calculation is performed on the heterogeneous dataset, taking into account the high- (penumbra driven) and low- (PDD and scatter-driven) gradient dose spreading. The former is driven mostly by target dose and surface shape, while the latter adds the dependence on target volume. This benchmark dose is used to produce the "best possible sparing" FDVH for an OAR, and based on it, progressively more easily achievable FDVH curves can be estimated. Validation was performed using test cylindrical geometries as well as 10 clinical HN datasets. For HN, VMAT plans were prepared with objectives of covering the primary and the secondary (bilateral elective neck) PTVs while addressing only one OAR at a time, with the goal of maximum sparing. The OARs were each parotid, the larynx, and the inferior pharyngeal constrictor. The difference in mean OAR doses was computed for the achieved vs. FDVHs, and the shapes of those DVHs were compared by means of the Dice similarity coefficient (DSC).

RESULTS

For all individually optimized HN OARs (N = 38), the average DSC between the planned DVHs and the FDVHs was 0.961 ± 0.018 (95% CI 0.955-0.967), with the corresponding average of mean OAR dose differences of 1.8 ± 5.8% (CI -0.1-3.6%). For realistic plans the achieved DVHs run no lower than the FDVHs, except when target coverage is compromised at the target/OAR interface.

CONCLUSIONS

For the validation of VMAT plans, the OAR DVHs optimized one-at-a-time were similar in shape to and bound on the low side by the FDVHs, within the confines of planner's ability to precisely cover the target(s) with the prescription dose(s). The method is best suited for the OARs close to the target. This approach is fundamentally different from "knowledge-based planning" because it is (a) independent of the treatment plan and prior experience, and (b) it approximates, from nearly first principles, the lowest possible boundary of the OAR DVH, but not necessarily its actual shape in the presence of competing OAR sparing and target dose homogeneity objectives.

Evaluation of a commercial automatic treatment planning system for prostate cancers

Recent developments in Radiation Oncology treatment planning have led to the development of software packages that facilitate automated intensity-modulated radiation therapy (IMRT) and volumetric-modulated arc therapy (VMAT) planning. Such solutions include site-specific modules, plan library methods, and algorithm-based methods. In this study, the plan quality for prostate cancer generated by the Auto-Planning module of the Pinnacle³ radiation therapy treatment planning system (v9.10, Fitchburg, WI) is retrospectively evaluated. The Auto-Planning module of Pinnacle³ uses a progressive optimization algorithm. Twenty-three prostate cancer cases, which had previously been planned and treated without lymph node irradiation, were replanned using the Auto-Planning module. Dose distributions were statistically compared with those of manual planning by the paired t-test at 5% significance level. Auto-Planning was performed without any manual intervention. Planning target volume (PTV) dose and dose to rectum were comparable between Auto-Planning and manual planning. The former, however, significantly reduced the dose to the bladder and femurs. Regression analysis was performed to examine the correlation between volume overlap between bladder and PTV divided by the total bladder volume and resultant V70. The findings showed that manual planning typically exhibits a logistic way for dose constraint, whereas Auto-Planning shows a more linear tendency. By calculating the Akaike information criterion (AIC) to validate the statistical model, a reduction of interoperator variation in Auto-Planning was shown. We showed that, for prostate cancer, the Auto-Planning module provided plans that are better than or comparable with those of manual planning. By comparing our results with those previously reported for head and neck cancer treatment, we recommend the homogeneous plan quality generated by the Auto-Planning module, which exhibits less dependence on anatomic complexity.

Iterative dataset optimization in automated planning: Implementation for breast and rectal cancer radiotherapy

PURPOSE

To develop a new automated treatment planning solution for breast and rectal cancer radiotherapy.

METHODS

The automated treatment planning solution developed in this study includes selection of the iterative optimized training dataset, dose volume histogram (DVH) prediction for the organs at risk (OARs), and automatic generation of clinically acceptable treatment plans. The iterative optimized training dataset is selected by an iterative optimization from 40 treatment plans for left-breast and rectal cancer patients who received radiation therapy. A two-dimensional kernel density estimation algorithm (noted as two parameters KDE) which incorporated two predictive features was implemented to produce the predicted DVHs. Finally, 10 additional new left-breast treatment plans are re-planned using the Pinnacle³ Auto-Planning (AP) module (version 9.10, Philips Medical Systems) with the objective functions derived from the predicted DVH curves. Automatically generated re-optimized treatment plans are compared with the original manually optimized plans.

RESULTS

By combining the iterative optimized training dataset methodology and two parameters KDE prediction algorithm, our proposed automated planning strategy improves the accuracy of the DVH prediction. The automatically generated treatment plans using the dose derived from the predicted DVHs can achieve better dose sparing for some OARs without compromising other metrics of plan quality.

CONCLUSIONS

The proposed new automated treatment planning solution can be used to efficiently evaluate and improve the quality and consistency of the treatment plans for intensity-modulated breast and rectal cancer radiation therapy.

Improving treatment plan evaluation with automation

The goal of this work is to evaluate the effectiveness of Plan‐Checker Tool (PCT) which was created to improve first‐time plan quality, reduce patient delays, increase the efficiency of our electronic workflow, and standardize and automate the physics plan review in the treatment planning system (TPS). PCT uses an application programming interface to check and compare data from the TPS and treatment management system (TMS). PCT includes a comprehensive checklist of automated and manual checks that are documented when performed by the user as part of a plan readiness check for treatment. Prior to and during PCT development, errors identified during the physics review and causes of patient treatment start delays were tracked to prioritize which checks should be automated. Nineteen of 33 checklist items were automated, with data extracted with PCT. There was a 60% reduction in the number of patient delays in the six months after PCT release. PCT was successfully implemented for use on all external beam treatment plans in our clinic. While the number of errors found during the physics check did not decrease, automation of checks increased visibility of errors during the physics check, which led to decreased patient delays. The methods used here can be applied to any TMS and TPS that allows queries of the database.

PACS number(s): 87.55.‐x, 87.55.N‐, 87.55.Qr, 87.55.tm, 89.20.Bb

Analytical modeling and feasibility study of a multi-GPU cloud-based server (MGCS) framework for non-voxel-based dose calculations

PURPOSE

In this paper, a multi-GPU cloud-based server (MGCS) framework is presented for dose calculations, exploring the feasibility of remote computing power for parallelization and acceleration of computationally and time intensive radiotherapy tasks in moving toward online adaptive therapies.

METHODS

An analytical model was developed to estimate theoretical MGCS performance acceleration and intelligently determine workload distribution. Numerical studies were performed with a computing setup of 14 GPUs distributed over 4 servers interconnected by a 1 Gigabits per second (Gbps) network. Inter-process communication methods were optimized to facilitate resource distribution and minimize data transfers over the server interconnect.

RESULTS

The analytically predicted computation time predicted matched experimentally observations within 1-5 %. MGCS performance approached a theoretical limit of acceleration proportional to the number of GPUs utilized when computational tasks far outweighed memory operations. The MGCS implementation reproduced ground-truth dose computations with negligible differences, by distributing the work among several processes and implemented optimization strategies.

CONCLUSIONS

The results showed that a cloud-based computation engine was a feasible solution for enabling clinics to make use of fast dose calculations for advanced treatment planning and adaptive radiotherapy. The cloud-based system was able to exceed the performance of a local machine even for optimized calculations, and provided significant acceleration for computationally intensive tasks. Such a framework can provide access to advanced technology and computational methods to many clinics, providing an avenue for standardization across institutions without the requirements of purchasing, maintaining, and continually updating hardware.

Initial evaluation of automated treatment planning software

Even with advanced inverse-planning techniques, radiation treatment plan opti-mization remains a very time-consuming task with great output variability, which prompted the development of more automated approaches. One commercially available technique mimics the actions of experienced human operators to pro-gressively guide the traditional optimization process with automatically created regions of interest and associated dose-volume objectives. We report on the initial evaluation of this algorithm on 10 challenging cases of locoreginally advanced head and neck cancer. All patients were treated with VMAT to 70 Gy to the gross disease and 56 Gy to the elective bilateral nodes. The results of post-treatment autoplanning (AP) were compared to the original human-driven plans (HDP). We used an objective scoring system based on defining a collection of specific dosimetric metrics and corresponding numeric score functions for each. Five AP techniques with different input dose goals were applied to all patients. The best of them averaged the composite score 8% lower than the HDP, across the patient population. The difference in median values was statistically significant at the 95% confidence level (Wilcoxon paired signed-rank test p = 0.027). This result reflects the premium the institution places on dose homogeneity, which was consistently higher with the HDPs. The OAR sparing was consistently better with the APs, the differences reaching statistical significance for the mean doses to the parotid glands (p < 0.001) and the inferior pharyngeal constrictor (p = 0.016), as well as for the maximum doses to the spinal cord (p = 0.018) and brainstem (p = 0.040). If one is prepared to accept less stringent dose homogeneity criteria from the RTOG 1016 protocol, nine APs would comply with the protocol, while providing lower OAR doses than the HDPs. Overall, AP is a promising clinical tool, but it could benefit from a better process for shifting the balance between the target dose coverage/homogeneity and OAR sparing.

Technology for Innovation in Radiation Oncology

Radiation therapy is an effective, personalized cancer treatment that has benefited from technological advances associated with the growing ability to identify and target tumors with accuracy and precision. Given that these advances have played a central role in the success of radiation therapy as a major component of comprehensive cancer care, the American Society for Radiation Oncology (ASTRO), the American Association of Physicists in Medicine (AAPM), and the National Cancer Institute (NCI) sponsored a workshop entitled "Technology for Innovation in Radiation Oncology," which took place at the National Institutes of Health (NIH) in Bethesda, Maryland, on June 13 and 14, 2013. The purpose of this workshop was to discuss emerging technology for the field and to recognize areas for greater research investment. Expert clinicians and scientists discussed innovative technology in radiation oncology, in particular as to how these technologies are being developed and translated to clinical practice in the face of current and future challenges and opportunities. Technologies encompassed topics in functional imaging, treatment devices, nanotechnology, and information technology. The technical, quality, and safety performance of these technologies were also considered. A major theme of the workshop was the growing importance of innovation in the domain of process automation and oncology informatics. The technologically advanced nature of radiation therapy treatments predisposes radiation oncology research teams to take on informatics research initiatives. In addition, the discussion on technology development was balanced with a parallel conversation regarding the need for evidence of efficacy and effectiveness. The linkage between the need for evidence and the efforts in informatics research was clearly identified as synergistic.

Can automation in radiotherapy reduce costs?

BACKGROUND

Computerized automation is likely to play an increasingly important role in radiotherapy. The objective of this study was to report the results of the first part of a program to implement a model for economical evaluation based on micro-costing method. To test the efficacy of the model, the financial impact of the introduction of an automation tool was estimated. A single- and multi-center validation of the model by a prospective collection of data is planned as the second step of the program.

MATERIAL AND METHODS

The model was implemented by using an interactive spreadsheet (Microsoft Excel, 2010). The variables to be included were identified across three components: productivity, staff, and equipment. To calculate staff requirements, the workflow of Gemelli ART center was mapped out and relevant workload measures were defined. Profit and loss, productivity and staffing were identified as significant outcomes. Results were presented in terms of earnings before interest and taxes (EBIT). Three different scenarios were hypothesized: baseline situation at Gemelli ART (scenario 1); reduction by 2 minutes of the average duration of treatment fractions (scenario 2); and increased incidence of advanced treatment modalities (scenario 3). By using the model, predicted EBIT values for each scenario were calculated across a period of eight years (from 2015 to 2022).

RESULTS

For both scenarios 2 and 3 costs are expected to slightly increase as compared to baseline situation that is particularly due to a little increase in clinical personnel costs. However, in both cases EBIT values are more favorable than baseline situation (EBIT values: scenario 1, 27%, scenario 2, 30%, scenario 3, 28% of revenues).

CONCLUSION

A model based on a micro-costing method was able to estimate the financial consequences of the introduction of an automation tool in our radiotherapy department. A prospective collection of data at Gemelli ART and in a consortium of centers is currently under way to prospectively validate the model.

Patient feature based dosimetric Pareto front prediction in esophageal cancer radiotherapy

PURPOSE

To investigate the feasibility of the dosimetric Pareto front (PF) prediction based on patient's anatomic and dosimetric parameters for esophageal cancer patients.

METHODS

Eighty esophagus patients in the authors' institution were enrolled in this study. A total of 2928 intensity-modulated radiotherapy plans were obtained and used to generate PF for each patient. On average, each patient had 36.6 plans. The anatomic and dosimetric features were extracted from these plans. The mean lung dose (MLD), mean heart dose (MHD), spinal cord max dose, and PTV homogeneity index were recorded for each plan. Principal component analysis was used to extract overlap volume histogram (OVH) features between PTV and other organs at risk. The full dataset was separated into two parts; a training dataset and a validation dataset. The prediction outcomes were the MHD and MLD. The spearman's rank correlation coefficient was used to evaluate the correlation between the anatomical features and dosimetric features. The stepwise multiple regression method was used to fit the PF. The cross validation method was used to evaluate the model.

RESULTS

With 1000 repetitions, the mean prediction error of the MHD was 469 cGy. The most correlated factor was the first principal components of the OVH between heart and PTV and the overlap between heart and PTV in Z-axis. The mean prediction error of the MLD was 284 cGy. The most correlated factors were the first principal components of the OVH between heart and PTV and the overlap between lung and PTV in Z-axis.

CONCLUSIONS

It is feasible to use patients' anatomic and dosimetric features to generate a predicted Pareto front. Additional samples and further studies are required improve the prediction model.

AutoLock: a semiautomated system for radiotherapy treatment plan quality control

A semiautomated system for radiotherapy treatment plan quality control (QC), named AutoLock, is presented. AutoLock is designed to augment treatment plan QC by automatically checking aspects of treatment plans that are well suited to computational evaluation, whilst summarizing more subjective aspects in the form of a checklist. The treatment plan must pass all automated checks and all checklist items must be acknowledged by the planner as correct before the plan is finalized. Thus AutoLock uniquely integrates automated treatment plan QC, an electronic checklist, and plan finalization. In addition to reducing the potential for the propagation of errors, the integration of AutoLock into the plan finalization workflow has improved efficiency at our center. Detailed audit data are presented, demonstrating that the treatment plan QC rejection rate fell by around a third following the clinical introduction of AutoLock.