Radiation Oncology Treatment Planning Research Articles

AAPM WGPE report 394: Simulated error training for the physics plan and chart review.

Simulated error training is a method to practice error detection in situations where the occurrence of error is low. Such is the case for the physics plan and chart review where a physicist may check several plans before encountering a significant problem. By simulating potentially hazardous errors, physicists can become familiar with how they manifest and learn from mistakes made during a simulated plan review. The purpose of this project was to develop a series of training datasets that allows medical physicists and trainees to practice plan and chart reviews in a way that is familiar and accessible, and to provide exposure to the various failure modes (FMs) encountered in clinical scenarios. A series of training datasets have been developed that include a variety of embedded errors based on the risk-assessment performed by American Association of Physicists in Medicine (AAPM) Task Group 275 for the physics plan and chart review. The training datasets comprise documentation, screen shots, and digital content derived from common treatment planning and radiation oncology information systems and are available via the Cloud-based platform ProKnow. Overall, 20 datasets have been created incorporating various software systems (Mosaiq, ARIA, Eclipse, RayStation, Pinnacle) and delivery techniques. A total of 110 errors representing 50 different FMs were embedded with the 20 datasets. The project was piloted at the 2021 AAPM Annual Meeting in a workshop where participants had the opportunity to review cases and answer survey questions related to errors they detected and their perception of the project's efficacy. In general, attendees detected higher-priority FMs at a higher rate, though no correlation was found between detection rate and the detectability of the FMs. Familiarity with a given system appeared to play a role in detecting errors, specifically when related to missing information at different locations within a given software system. Overall, 96% of respondents either agreed or strongly agreed that the ProKnow portal and training datasets were effective as a training tool, and 75% of respondents agreed or strongly agreed that they planned to use the tool at their local institution. The datasets and digital platform provide a standardized and accessible tool for training, performance assessment, and continuing education regarding the physics plan and chart review. Work is ongoing to expand the project to include more modalities, radiation oncology treatment planning and information systems, and FMs based on emerging techniques such as auto-contouring and auto-planning.

Assessing out-of-pocket cost differences in treatment for Medicare-eligible patients with favorable prognosis early-stage Hodgkin lymphoma.

15 Background: In the treatment of favorable Hodgkin Lymphoma (HL), Level 1 evidence comparing combined modality therapy (CMT) with chemotherapy alone has persistently demonstrated that obviating radiation therapy (RT) worsens recurrence rates despite no significant overall survival difference. As there remains limited evidence on the granularity of the impact of insurance on HL treatment, this project disaggregates costs by insurance plans to enhance the transparency of out-of-pocket (OOP) cost estimations in Medicaid and/or Medicare-eligible (age 65+) patients. Methods: National Comprehensive Cancer Network guidelines were queried to assess the appropriate therapy for HL. Following initial adriamycin/bleomycin/vinblastine/dacarbazine (ABVD) chemotherapy x 2 cycles with Deauville 1-2 response on PET scan, CMT was defined as involved site RT (20 Gy / 10 fractions), while chemotherapy alone was defined as an additional ABVD x 2. Current Procedural Terminology and Healthcare Common Procedure Coding System codes identified treatments. OOP expenditures, deductibles, and copay/coinsurance were computed for Medicaid, Original Medicare, and Medigap Plan G plans via medicare.gov, medicaid.oh.gov, and cms.gov. Price estimates (from actual prices per insurance plan rather than claims data predictions) assume an Ohio hospital location (zip code = 44106) for all treatments. Total costs were calculated based on all treatments plus a 5-year follow-up (not adjusted for inflation). Treatment charges for chemotherapy alone include initial medical oncology office visit, cost of ABVD x 4, and infusion costs for 4 cycles. Additional treatment charges for CMT include initial radiation oncology consultation, treatment planning, dosimetry calculations, treatment device, simulation and verification, RT delivery, on-treatment visits, medical physics consultation, and annual TSH labs. Results: With Original Medicare plans, patients are faced with an OOP cost of 20% for Medicare Part B claims with no cost cap for approved procedures after the deductible. As such, patients are faced with a total 6-year charge of $1,928.95 for chemotherapy alone, and $2,385.28 for CMT. Under Medigap Plan G, patients encounter an OOP charge of $1,151.14 for chemotherapy alone and $1,176.34 for CMT. Medicaid beneficiaries (assuming all treatments are approved by Medicaid) face no OOP expenses for both treatment plans, as all expenses are covered without limit. Conclusions: By comparing OOP costs between Medicare and Medicaid plans, this model improves cost transparency in the treatment of favorable early-stage HL. Our findings indicate that omitting RT in favor of chemotherapy alone yields a cost savings of 2-24% at a 5-year follow-up. Further studies are required to assess whether these savings compensate for the inferior outcomes achieved by omitting RT in this patient population.

Evaluation of Peer Review Practice in Radiation Oncology Treatment Planning: Single Institution Experience

Pretreatment Peer review meeting at our radiation oncology department is a weekly meeting, where the initial radiation treatment (RT) planning process for every patient is discussed including: Computed Tomography (CT) simulation and treatment planning. In this study, we are aiming to evaluate our peer review process and to report the compliance to the suggested modifications. All RT plans for all patients indicated for radiotherapy accepted by the primary radiation oncology team then presented in our weekly peer review meeting for final approval. A minimum of two radiation oncologists reviewed each plan and their suggestions documented then suggested modifications are classified as changes in: field arrangement, clinical target volume (CTV), dose and fractionation. Compliance to suggested modification also recorded as: applied and not applied with or without documented justification. Atotal of 1779 radiation treatment plans were discussed in our regular weekly peer review meetings for a period of 3 years. 67.5% of them were treated with radical intent while 32.5% treated with palliative intent. The most common technique used was three-dimensional conformal (3DCRT) followed by intensity modulated RT (IMRT) (70.6% and 29.4%, respectively). The most frequently reviewed sites were: breast (38%), Gastrointestinal (11.9%) and head and neck (9.3%).1592 plans (89.48%) were labeled as: approved, 162 (9.1%) labeled as: modified and only 25 (1.4%) plans labeled as: hold. The frequency of modifications was significantly greater in breast (39.5%) than Gastrointestinal (17.9%) followed by gynecological cancers (14.2%) (P value <0.001). Modifications were more predominant with radical plans (64.2%) than palliative ones (35.8%) (P value <0.001) and statistically greater using 3DCRT (64.2%) than IMRT (35.8%) (P value <0.001). CTV recorded as the most frequent variable suggested for modifications (34.6%), followed by dose and fractionation (24.7%) (P value = 0.035). 72.2% of recommended modifications were applied by the consultant of primary team and 22.2% of non-applied modifications were justified and documented. There was 100% compliance rate to all plans labeled as hold. The practiceof peer review in radiation oncology specialty is very challenging. Its role in quality assurance of the radiotherapy process is highly accepted by radiation oncologists at our radiation oncology department. It is reflected in their high compliance rate to the suggested modifications. We recommend its implementation routinely at any radiation oncology department to enhance the quality of RT plans.

Clinical Outcomes of Image-Guided Volumetric Modulated Arc Therapy for Total Body Irradiation

Volumetric modulated arc therapy (VMAT)-based total body irradiation (TBI) with image guidance is a novel technique that is increasing in implementation. Compared to conventional TBI, VMAT-TBI offers favorable dose homogeneity, better organ-at-risk sparing, and enhanced patient comfort. However, whether these dosimetric advantages translate to improved clinical outcomes that justify the increased planning and delivery burden is not well understood. Only a single study of clinical outcomes of VMAT-TBI exists in the literature. We present the largest study to date of clinical outcomes of VMAT-TBI. In this IRB-approved retrospective single-institution study, all patients treated with VMAT-TBI conditioning for allogeneic stem cell transplant, per the institution's published protocol, were identified. Dosimetric data were abstracted from the radiation oncology treatment planning system. Clinical data were abstracted from the electronic medical record. The primary outcome was six-month overall survival (6M OS) from the last day of TBI by Kaplan-Meier method. Fifty-five patients (47 adult and 8 pediatric) were treated with VMAT-TBI between June 2020 and December 2022. All patients received conditioning chemotherapy with standard-dose TBI of 12 or 13.2 Gy in 8 twice-daily fractions. The PTV coverage (V95%) mean was 95.3% ± 1.2%. Mean lung dose was 9.5 Gy ± 0.6 for adult patients and 8.4 Gy ± 0.9 for pediatric patients. Mean lung dose rate was 18.0 cGy/min ± 4.4. Mean kidney dose was 5.9 Gy ± 0.6. Mean skin dose measured by MOSFET was 12.7 Gy ± 1.2. Median treatment time was 63 minutes (range: 53-104). Median follow-up was 7.7 months. At most recent follow-up, 78% of patients were alive. 6M OS was 82%. Common acute toxicities were fatigue (90.9% of patients, all grade 1-2), diarrhea (70.9%, all grade 1-2), nausea (76.4%, all grade 1-2), mucositis (60% grade 1-2, 12.7% grade 3, 1.8% grade 4, no grade 5), and xerostomia (54.5%, all grade 1). Mean pretreatment FEV1 was 98.3 percent of predicted (%p) ± 11.9%p and mean posttreatment FEV1 was 94.7%p ± 13.8%p. Mean pretreatment GFR was 101.4mL/min/1.73m² ± 17.4, mean 3-month posttreatment GFR was 92.4 ± 20.0, and mean 6-month posttreatment GFR was 97.5 ± 26.48. One patient experienced grade 2 pneumonitis; there were no other cases of pneumonitis. There were no acute grade 3+ toxicities aside from mucositis. Observed late toxicities were cataracts (7.3%, all grades 1-3) and hypothyroidism (12.7%, all grades 1-2). There were no grade 3+ late toxicities. Mild acute graft-versus-host disease (GVHD) was noted in 27.2% of patients and mild chronic GVHD was noted in 14.5% of patients, with no other cases of GVHD. In the largest series to date, VMAT-TBI had excellent oncologic and toxicity outcomes. A randomized trial of VMAT-TBI versus standard TBI is warranted.

Automated Segmentation of Sacral Chordoma and Surrounding Muscles Using Deep Learning Ensemble

The manual segmentation of organ structures in radiation oncology treatment planning is a time-consuming and highly skilled task, particularly when treating rare tumors like sacral chordomas. This study evaluates the performance of automated deep learning (DL) models in accurately segmenting the gross tumor volume (GTV) and surrounding muscle structures of sacral chordomas. An expert radiation oncologist contoured 5 muscle structures (gluteus maximus, gluteus medius, gluteus minimus, paraspinal, piriformis) and sacral chordoma GTV on computed tomography images from 48 patients. We trained 6 DL auto-segmentation models based on 3-dimensional U-Net and residual 3-dimensional U-Net architectures. We then implemented an average and an optimally weighted average ensemble to improve prediction performance. We evaluated algorithms with the average and standard deviation of the volumetric Dice similarity coefficient, surface Dice similarity coefficient with 2- and 3-mm thresholds, and average symmetric surface distance. One independent expert radiation oncologist assessed the clinical viability of the DL contours and determined the necessary amount of editing before they could be used in clinical practice. Quantitatively, the ensembles performed the best across all structures. The optimal ensemble (volumetric Dice similarity coefficient, average symmetric surface distance) was (85.5 ± 6.4, 2.6 ± 0.8; GTV), (94.4 ± 1.5, 1.0 ± 0.4; gluteus maximus), (92.6 ± 0.9, 0.9 ± 0.1; gluteus medius), (85.0 ± 2.7, 1.1 ± 0.3; gluteus minimus), (92.1 ± 1.5, 0.8 ± 0.2; paraspinal), and (78.3 ± 5.7, 1.5 ± 0.6; piriformis). The qualitative evaluation suggested that the best model could reduce the total muscle and tumor delineation time to a 19-minute average. Our methodology produces expert-level muscle and sacral chordoma tumor segmentation using DL and ensemble modeling. It can substantially augment the streamlining and accuracy of treatment planning and represents a critical step toward automated delineation of the clinical target volume in sarcoma and other disease sites.

Feasibility of a multigroup Boltzmann–Fokker–Planck solution for electron beam dose calculations

Legacy nuclear-reactor Boltzmann solvers start clinical deployment as an alternative to Monte Carlo (MC) codes and Fermi–Eyges semiemprical models in radiation oncology treatment planning. Today’s certified clinical solvers are limited to photon beams. In this paper, ELECTR, a state-of-the-art multigroup electron cross sections generation module in NJOY is presented and validated against Lockwood’s calorimetric measurements, EGS-nrc and GEANT-4 for 1–20 MeV unidirectional electron beams. The nuclear-reactor DRAGON-5 solver is upgraded to access the library and solve the Boltzmann–Fokker–Planck (BFP) equation. A variety of heterogeneous radiotherapy and radiosurgery phantom configurations were used for validation purpose. Case studies include a thorax benchmark, that of a typical breast Intra-Operative Radiotherapy and a high-heterogeneity patient-like benchmark. For all beams, 100% of the water voxels satisfied the American Association of Physicists in Medicine accuracy criterion for a BFP-MC dose error below 2%. At least, 97.0% of adipose, muscle, bone, lung, tumor and breast voxels satisfied the 2% criterion. The average BFP-MC relative error was about 0.56% for all voxels, beams and materials combined. By irradiating homogeneous slabs from Z=1 (hydrogen) to Z=99 (einsteinium), we reported performance and defects of the CEPXS mode [US. Sandia National Lab., SAND-89-1685] in ELECTR for the entire periodic table. For all Lockwood’s benchmarks, NJOY-DRAGON dose predictions are within the experimental data precision for 98% of voxels.

The importance of standardized treatment planning and decision-making in radiation oncology for non-small-cell lung cancer—are current guidelines sufficiently strict for uniform target delineation?—a narrative literature review

The importance of standardized treatment planning and decision-making in radiation oncology for non-small-cell lung cancer—are current guidelines sufficiently strict for uniform target delineation?—a narrative literature review

Systematic Implementation of Effective Quality Assurance Processes for the Assessment of Radiation Target Volumes in Head and Neck Cancer

<h3>Purpose/Objective(s)</h3> Target volume delineation and treatment planning in Radiation Oncology for head and neck cancer (HNC) cases can be challenging and time consuming. Volume delineation requires significant expertise and is subject to significant inter-physician variability, and changes in treatment volumes after planning has started are associated with planning delays. While chart rounds are a major quality assurance (QA) intervention for treatment planning, few strategies have been developed to effectively evaluate volume delineation. The aim of the study is to evaluate the feasibility, effectiveness and the potential clinical impact of implementation of prospective Quality Assurance (QA) Processes for the assessment of radiation target volumes. <h3>Materials/Methods</h3> Starting in 2020, a systematic weekly QA process with real-time metrics was adopted in our department for the assessment of HNC planning target volumes (PTVs) before the start of radiation planning. All definitive-intent treatment cases (intact and post-operative) were evaluated. Contours were flagged with green (none), yellow (minor), or red colors (major) based on the degree of required modifications. <h3>Results</h3> 400 consecutive HNC cases were included. A total of 37 (9%) PTVs required modifications, 5 (1%) of these were flagged as red (major) deemed highly likely to impact oncologic outcomes – for example, omission of gross disease and/or elective areas at high risk of microscopic disease. Thirty-two (8%) were flagged yellow (minor) deemed potentially to impact oncologic or functional outcomes—for example, omission of elective areas at intermediate or low risk of harboring microscopic disease, or errors of over-contouring volumes into normal critical structures. All contouring modifications were performed before the start of radiation planning, resulting in significant reduction in the amount of re-planning by our dosimetry group and zero delays in treatment start due to planning within this cohort. Interestingly, between the first 200 and second 200 HNC cases reviewed, major modifications decreased from 5 to 0 and minor modifications from 22 to 10 (chi squared p=0.02), suggesting the presence of a learning curve and standardization of practice of the treating physicians even at a high-volume academic center. <h3>Conclusion</h3> Our study highlights the feasibility and importance of systematic contour QA in head and neck radiation oncology. Further evaluation and implementation of these efforts across other disease sites appear warranted.

Harmonisation of scanner-dependent contrast variations in magnetic resonance imaging for radiation oncology, using style-blind auto-encoders.

Magnetic Resonance Imaging (MRI) exhibits scanner dependent contrast, which limits generalisability of radiomics and machine-learning for radiation oncology. Current deep-learning harmonisation requires paired data, retraining for new scanners and often suffers from geometry-shift which alters anatomical information. The aim of this study was to investigate style-blind auto-encoders for MRI harmonisation to accommodate unpaired training data, avoid geometry-shift and harmonise data from previously unseen scanners. A style-blind auto-encoder, using adversarial classification on the latent-space, was designed for MRI harmonisation. The public CC359 T1-w MRI brain dataset includes six scanners (three manufacturers, two field strengths), of which five were used for training. MRI from all six (including one unseen) scanner were harmonised to common contrast. Harmonisation extent was quantified via Kolmogorov-Smirnov testing of residual scanner dependence of 3D radiomic features, and compared to WhiteStripe normalisation. Anatomical content preservation was measured through change in structural similarity index on contrast-cycling (δSSIM). The percentage of radiomics features showing statistically significant scanner-dependence was reduced from 41% (WhiteStripe) to 16% for white matter and from 39% to 27% for grey matter. δSSIM<0.0025 on harmonisation and de-harmonisation indicated excellent anatomical content preservation. Our method harmonised MRI contrast effectively, preserved critical anatomical details at high fidelity, trained on unpaired data and allowed zero-shot harmonisation. Robust and clinically translatable harmonisation of MRI will enable generalisable radiomic and deep-learning models for a range of applications, including radiation oncology treatment stratification, planning and response monitoring.

National reference levels of CT procedures dedicated for treatment planning in radiation oncology

ObjectiveTo present results of the first national survey on reference levels of CT imaging performed for the treatment planning purposes in radiation oncology in Croatia. MethodsData for CT protocols of five anatomical regions including head, head and neck, pelvis, breast, and thorax were collected at eight radiation oncology departments in Croatia. Data included volume CT dose index (CTDIvol), dose-length product (DLP), scan length and set of acquisition and reconstruction parameters. Data on a total of 600 patients were collected. Median values of scan length, DLP and CTDIvol were calculated for each acquisition protocol. Third quartiles of the median CTDIvol and DLP values were proposed as the national radiotherapy planning reference levels (RPRL). ResultsThe largest CoV were assessed for RT Breast (63.8% for CTDIvol), RT Thorax (79.7% for DLP) and RT H&N (21.2% for scan length). RT Head had the lowest CoV for CTDIvol (1,9%) and DLP (17,2%), while RT Breast had the lowest coefficient of variation for scan length (12.8%). Proposed national RPRLs are: for RT Head CTDIvol16cm = 62 mGy and DLP16cm = 1738 mGy.cm; for RT H&N CTDIvol16cm = 35 mGy and DLP16cm = 1444 mGy.cm; for RT Breast CTDIvol32cm = 16 mGy and DLP32cm = 731 mGy.cm; for RT Thorax CTDIvol32cm = 17 mGy and DLP32cm = 865 mGy.cm; for RT Pelvis CTDIvol32cm = 20 mGy and DLP32cm = 1133 mGy.cm. ConclusionsResults of this study show variations in CT imaging for treatment planning practice at the national level which call for optimization of procedures.

Using Incident Reports to Advance Patient Safety and System Resilience in Radiation Oncology Treatment Planning

We used incident reports to improve treatment planning in radiation oncology to advance patient safety. First, a cluster of related reports enabled discovery of communication issues for setup instructions stemming from a text box with a limited character size. A redesigned process used structured data and image formats to reduce the proportion of inaccurate setup instructions. An upward trend in inaccurate setup instructions identified that some rotating staff were unaware of which equipment to use. Implementing default disease site setups ensured that when a patient was to commence treatment for a specific disease site, it was known which immobilizing equipment should be utilized as the basis before customization. Second, inappropriate plans were identified for proactive improvement. The process resilience was enhanced with targeted validations of generated treatment plans. The automated checks incorporated data automatically pulled from the electronic health record. Useful methods included clustering reports into categories to proactively see patterns in reported issues, collaborating with clinical experts from different perspectives, surveillance monitoring to identify new risks as conditions change, improving communications across disciplines through health information technology, and increasing system resilience with targeted validation of known gaps and variation in plans. The implementation of ILS frameworks, careful review of events and categorization, which culminate with interventions targeting the most frequent or highest severity incidents reported, increased patient safety within our department.

Deep-learning system to improve the quality and efficiency of volumetric heart segmentation for breast cancer

Although artificial intelligence algorithms are often developed and applied for narrow tasks, their implementation in other medical settings could help to improve patient care. Here we assess whether a deep-learning system for volumetric heart segmentation on computed tomography (CT) scans developed in cardiovascular radiology can optimize treatment planning in radiation oncology. The system was trained using multi-center data (n = 858) with manual heart segmentations provided by cardiovascular radiologists. Validation of the system was performed in an independent real-world dataset of 5677 breast cancer patients treated with radiation therapy at the Dana-Farber/Brigham and Women’s Cancer Center between 2008–2018. In a subset of 20 patients, the performance of the system was compared to eight radiation oncology experts by assessing segmentation time, agreement between experts, and accuracy with and without deep-learning assistance. To compare the performance to segmentations used in the clinic, concordance and failures (defined as Dice < 0.85) of the system were evaluated in the entire dataset. The system was successfully applied without retraining. With deep-learning assistance, segmentation time significantly decreased (4.0 min [IQR 3.1–5.0] vs. 2.0 min [IQR 1.3–3.5]; p < 0.001), and agreement increased (Dice 0.95 [IQR = 0.02]; vs. 0.97 [IQR = 0.02], p < 0.001). Expert accuracy was similar with and without deep-learning assistance (Dice 0.92 [IQR = 0.02] vs. 0.92 [IQR = 0.02]; p = 0.48), and not significantly different from deep-learning-only segmentations (Dice 0.92 [IQR = 0.02]; p ≥ 0.1). In comparison to real-world data, the system showed high concordance (Dice 0.89 [IQR = 0.06]) across 5677 patients and a significantly lower failure rate (p < 0.001). These results suggest that deep-learning algorithms can successfully be applied across medical specialties and improve clinical care beyond the original field of interest.

RTGAN: ORGAN CONTOURS IN RADIATION THERAPY WITH GENERATIVE ADVERSARIAL NETWORK

Generally, radiation oncology applies evaluation and prediction in medical imaging and diagnosis, specifically for contouring organs, which results in the production of the clinical target volume (CTV) that corresponds to disease risk and organ exclusion. Medical physicists contour organs and combine computed tomography (CT) scans to digital imaging and communications in medicine (DICOM) radiation therapy (RT) to assist physicians for diagnosing tumors and calculating the dosages in treatments including radiation and chemotherapy. Thus, to generate RT images with high accuracy, this paper proposes a new Generator Adversarial Network (GAN) for RT images called radiation therapy GAN (RTGAN). We combine multiple loss functions with synthetic similarity DICOM-RT images and compare the results with Pinnacle, a radiation oncology treatment planning system. Further, we evaluate the method to get a score of 0.984 in structured similarity (SSIM) and 31.26 in peak signal-to-noise ratio (PSNR) and find that it costs 0.058 s to finish contouring one CT image. The proposed method is applied and tested in the department of radiation oncology at the Chung Shan Medical University Hospital, and the results are similar to the ground truth images. Thus, it not only effectively reduces the false-positive rate but also makes a breakthrough in medicine.

Abstract IA-22: Automated treatment planning and quality assurance in radiation oncology

Abstract The delivery of radiation for the treatment of cancer is a complicated process that requires both clinical and technical expertise to ensure radiation treatments are safe and effective. Sub-optimal radiation treatments have the potential to result in significant detriment to the patient and several studies have shown radiation treatments, which deviate from established clinical guidelines, result in worse patient outcomes. Therefore, the current radiation treatment process requires substantial multi-disciplinary resources to both generate and verify radiation treatments are of high-quality. In this talk, a previously validated machine learning platform customized for radiation oncology will be presented. The method automatically learns based on data from thousands of previously treated patients which relationships and patterns in radiation oncology image and treatment data and has been applied for automated data mining activities, automated quality assurance to support expedited radiation treatment review, and for radiation dose prediction to develop new radiation treatments by best deciding where dose should be placed and how dose should be delivered without requiring any manual intervention. Therefore, the method can be used to both generate personalized radiation treatments and to quantitatively score radiation treatments for quality and classify radiation treatments that have errors. The automated platform can readily be integrated into current clinical process to improve efficiency in the radiation treatment planning and plan review processes and to better utilize the vast data we have to ensure we are providing patients with highly personalized radiation treatments. Citation Format: Thomas G. Purdie. Automated treatment planning and quality assurance in radiation oncology [abstract]. In: Proceedings of the AACR Virtual Special Conference on Artificial Intelligence, Diagnosis, and Imaging; 2021 Jan 13-14. Philadelphia (PA): AACR; Clin Cancer Res 2021;27(5_Suppl):Abstract nr IA-22.

Rigid and Deformable Image Registration for Radiation Therapy: A Self-Study Evaluation Guide for NRG Oncology Clinical Trial Participation

PurposeThe registration of multiple imaging studies to radiation therapy computed tomography simulation, including magnetic resonance imaging, positron emission tomography-computed tomography, etc. is a widely used strategy in radiation oncology treatment planning, and these registrations have valuable roles in image guidance, dose composition/accumulation, and treatment delivery adaptation. The NRG Oncology Medical Physics subcommittee formed a working group to investigate feasible workflows for a self-study credentialing process of image registration commissioning. Methods and MaterialsThe American Association of Physicists in Medicine (AAPM) Task Group 132 (TG132) report on the use of image registration and fusion algorithms in radiation therapy provides basic guidelines for quality assurance and quality control of the image registration algorithms and the overall clinical process. The report recommends a series of tests and the corresponding metrics that should be evaluated and reported during commissioning and routine quality assurance, as well as a set of recommendations for vendors. The NRG Oncology medical physics subcommittee working group found incompatibility of some digital phantoms with commercial systems. Thus, there is still a need to provide further recommendations in terms of compatible digital phantoms, clinical feasible workflow, and achievable thresholds, especially for future clinical trials involving deformable image registration algorithms. Nine institutions participated and evaluated 4 commonly used commercial imaging registration software and various versions in the field of radiation oncology. Results and ConclusionsThe NRG Oncology Working Group on image registration commissioning herein provides recommendations on the use of digital phantom/data sets and analytical software access for institutions and clinics to perform their own self-study evaluation of commercial imaging systems that might be employed for coregistration in radiation therapy treatment planning and image guidance procedures. Evaluation metrics and their corresponding values were given as guidelines to establish practical tolerances. Vendor compliance for image registration commissioning was evaluated, and recommendations were given for future development.

Validation of irtGPUMCD, a GPU-based Monte Carlo internal dosimetry framework for radionuclide therapy

PurposeMonte Carlo (MC) simulations are highly desirable for dose treatment planning and evaluation in radiation oncology. This is true also in emerging nuclear medicine applications such as internal radiotherapy with radionuclides. The purpose of this study is the validation of irtGPUMCD, a GPU-based MC code for dose calculations in internal radiotherapy. MethodsThe female and male phantoms of the International Commission on Radiological Protection (ICRP 110) were used as benchmarking geometries for this study focused on 177Lu and including 99mTc and 131I. Dose calculations were also conducted for a real patient. For phantoms, twelve anatomical structures were considered as target/source organs. The S-values were evaluated with irtGPUMCD simulations (108 photons), with gamma branching ratios of ICRP 107 publication. The 177Lu electrons S-values were calculated for source organs only, based on local deposition of dose in irtGPUMCD. The S-value relative difference between irtGPUMCD and IDAC-DOSE were evaluated for all targets/sources considered. A DVHs comparison with GATE was conducted. An exponential track length estimator was introduced in irtGPUMCD to increase computational efficiency. ResultsThe relative S-value differences between irtGPUMCD and IDAC-DOSE were <5% while this comparison with GATE was <1%. The DVHs dosimetric indices comparison between GATE and irtGPUMCD for the patient led to an excellent agreement (<2%). The time required for the simulation of 108 photons was 1.5 min for the female phantom, and one minute for the real patient (<1% uncertainty). These results are promising and let envision the use of irtGPUMCD for internal dosimetry in clinical applications.

Artificial intelligence in radiation oncology treatment planning: a brief overview

Among medical specialties, radiation oncology has long been an innovator and early adopter of therapeutic technologies. This specialty is now situated in prime position to be revolutionized by advances in artificial intelligence (AI), especially machine and deep learning. AI has been investigated by radiation oncologists and physicists in both general and niche radiotherapy planning tasks and has often demonstrated performance that is indistinguishable from human experts, while substantially shortening the time required to complete these tasks. We sought to review applications of AI to domains germane to radiation oncology, namely: image segmentation, treatment plan generation and optimization, normal tissue complication probability modeling, quality assurance (QA), and adaptive re-planning. We sought likewise to consider obstacles to AI adoption in the radiotherapy clinic, now primarily political, legal, and ethical rather than technical in nature.

Quantifying potential reduction in contrast dose with monoenergetic images synthesized from dual-layer detector spectral CT.

To estimate the potential dose reduction in iodinated contrast when interpreting monoenergetic images from spectral CT. 51 paediatric patients received contrast-enhanced CT simulation for radiation therapy using a single-source, dual-layer detector spectral CT. The contrast-to-noise ratios (CNRs) of blood vessels were measured relative to surrounding soft tissue. CNRs on monoenergetic 40-70 keV images were compared with polychromatic 120 kVp images. To compare with in vivo results, a phantom with iodine inserts (2-20 mg ml-1 concentration) was scanned and CNRs were calculated relative to water background. Monoenergetic keV and body site had significant effects on CNR ratio (p < 0.0001). Across all body sites, the mean CNR ratio (monoenergetic/polychromatic CNR) was 3.3 (20th percentile [%20] 2.6), 2.4 (%20 2.1), 1.7 (%20 1.5), 1.2 (%20 1.0) for 40, 50, 60 and 70 keV images, respectively. Image noise was highest at 40 keV and lowest at 70 keV. Phantom measurements indicated that the same CNR as 120 kVp images can be achieved with a 4.0-fold lower iodine concentration on 40 keV images and 2.5-fold lower on 50 keV images. 50 keV monoenergetic images provided the best balance of improved CNR on all studies (mean 2.4-fold increase in vivo) for enhancing vessels vs image noise. A 50% reduction in contrast dose on a 50 keV image should maintain comparable or better CNR as compared with polychromatic CT in over 80% of CT studies. Advances in knowledge: Use of a novel, single-source, dual-layer detector spectral CT scanner to improve visualization of contrast-enhanced blood vessels will reduce the amount of iodinated contrast required for radiation oncology treatment planning.

Investigating multi-objective fluence and beam orientation IMRT optimization

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

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.