K Moore

EXPERIENCE-BASED QUALITY CONTROL OF CLINICAL INTENSITY-MODULATED RADIOTHERAPY PLANNING

PURPOSE To incorporate a quality control tool, according to previous planning experience and patient-specific anatomic information, into the intensity-modulated radiotherapy (IMRT) plan generation process and to determine whether the tool improved treatment plan quality. METHODS AND MATERIALS A retrospective study of 42 IMRT plans demonstrated a correlation between the fraction of organs at risk (OARs) overlapping the planning target volume and the mean dose. This yielded a model, predicted dose = prescription dose (0.2 + 0.8 [1 - exp(-3 overlapping planning target volume/volume of OAR)]), that predicted the achievable mean doses according to the planning target volume overlap/volume of OAR and the prescription dose. The model was incorporated into the planning process by way of a user-executable script that reported the predicted dose for any OAR. The script was introduced to clinicians engaged in IMRT planning and deployed thereafter. The scripts effect was evaluated by tracking δ = (mean dose-predicted dose)/predicted dose, the fraction by which the mean dose exceeded the model. RESULTS All OARs under investigation (rectum and bladder in prostate cancer; parotid glands, esophagus, and larynx in head-and-neck cancer) exhibited both smaller δ and reduced variability after script implementation. These effects were substantial for the parotid glands, for which the previous δ = 0.28 ± 0.24 was reduced to δ = 0.13 ± 0.10. The clinical relevance was most evident in the subset of cases in which the parotid glands were potentially salvageable (predicted dose <30 Gy). Before script implementation, an average of 30.1 Gy was delivered to the salvageable cases, with an average predicted dose of 20.3 Gy. After implementation, an average of 18.7 Gy was delivered to salvageable cases, with an average predicted dose of 17.2 Gy. In the prostate cases, the rectum model excess was reduced from δ = 0.28 ± 0.20 to δ = 0.07 ± 0.15. On surveying dosimetrists at the end of the study, most reported that the script both improved their IMRT planning (8 of 10) and increased their efficiency (6 of 10). CONCLUSIONS This tool proved successful in increasing normal tissue sparing and reducing interclinician variability, providing effective quality control of the IMRT plan development process.

Cloud computing in medical imaging

Over the past century technology has played a decisive role in defining, driving, and reinventing procedures, devices, and pharmaceuticals in healthcare. Cloud computing has been introduced only recently but is already one of the major topics of discussion in research and clinical settings. The provision of extensive, easily accessible, and reconfigurable resources such as virtual systems, platforms, and applications with low service cost has caught the attention of many researchers and clinicians. Healthcare researchers are moving their efforts to the cloud, because they need adequate resources to process, store, exchange, and use large quantities of medical data. This Vision 20/20 paper addresses major questions related to the applicability of advanced cloud computing in medical imaging. The paper also considers security and ethical issues that accompany cloud computing.

Knowledge‐based prediction of plan quality metrics in intracranial stereotactic radiosurgery

PURPOSE The objective of this work was to develop a comprehensive knowledge-based methodology for predicting achievable dose-volume histograms (DVHs) and highly precise DVH-based quality metrics (QMs) in stereotactic radiosurgery/radiotherapy (SRS/SRT) plans. Accurate QM estimation can identify suboptimal treatment plans and provide target optimization objectives to standardize and improve treatment planning. METHODS Correlating observed dose as it relates to the geometric relationship of organs-at-risk (OARs) to planning target volumes (PTVs) yields mathematical models to predict achievable DVHs. In SRS, DVH-based QMs such as brain V10Gy (volume receiving 10 Gy or more), gradient measure (GM), and conformity index (CI) are used to evaluate plan quality. This study encompasses 223 linear accelerator-based SRS/SRT treatment plans (SRS plans) using volumetric-modulated arc therapy (VMAT), representing 95% of the institutions VMAT radiosurgery load from the past four and a half years. Unfiltered models that use all available plans for the model training were built for each category with a stratification scheme based on target and OAR characteristics determined emergently through initial modeling process. Model predictive accuracy is measured by the mean and standard deviation of the difference between clinical and predicted QMs, δQM = QMclin - QMpred, and a coefficient of determination, R(2). For categories with a large number of plans, refined models are constructed by automatic elimination of suspected suboptimal plans from the training set. Using the refined model as a presumed achievable standard, potentially suboptimal plans are identified. Predictions of QM improvement are validated via standardized replanning of 20 suspected suboptimal plans based on dosimetric predictions. The significance of the QM improvement is evaluated using the Wilcoxon signed rank test. RESULTS The most accurate predictions are obtained when plans are stratified based on proximity to OARs and their PTV volume sizes. Volumes are categorized into small (VPTV < 2 cm(3)), medium (2 cm(3) < VPTV < 25 cm(3)), and large (25 cm(3) < VPTV). The unfiltered models demonstrate the ability to predict GMs to ∼1 mm and fractional brain V10Gy to ∼25% for plans with large VPTV and critical OAR involvements. Increased accuracy and precision of QM predictions are obtained when high quality plans are selected for the model training. For the small and medium VPTV plans without critical OAR involvement, predictive ability was evaluated using the refined model. For training plans, the model predicted GM to an accuracy of 0.2 ± 0.3 mm and fractional brain V10Gy to 0.04 ± 0.12, suggesting highly accurate predictive ability. For excluded plans, the average δGM was 1.1 mm and fractional brain V10Gy was 0.20. These δQM are significantly greater than those of the model training plans (p < 0.001). For CI, predictions are close to clinical values and no significant difference was observed between the training and excluded plans (p = 0.19). Twenty outliers with δGM > 1.35 mm were identified as potentially suboptimal, and replanning these cases using predicted target objectives demonstrates significant improvements on QMs: on average, 1.1 mm reduction in GM (p < 0.001) and 23% reduction in brain V10Gy (p < 0.001). After replanning, the difference of δGM distribution between the 20 replans and the refined model training plans was marginal. CONCLUSIONS The results demonstrate the ability to predict SRS QMs precisely and to identify suboptimal plans. Furthermore, the knowledge-based DVH predictions were directly used as target optimization objectives and allowed a standardized planning process that bettered the clinically approved plans. Full clinical application of this methodology can improve consistency of SRS plan quality in a wide range of PTV volume and proximity to OARs and facilitate automated treatment planning for this critical treatment site.

Automated radiotherapy treatment plan integrity verification

PURPOSE In our clinic, physicists spend from 15 to 60 min to verify the physical and dosimetric integrity of radiotherapy plans before presentation to radiation oncology physicians for approval. The purpose of this study was to design and implement a framework to automate as many elements of this quality control (QC) step as possible. METHODS A comprehensive computer application was developed to carry out a majority of these verification tasks in the Philips PINNACLE treatment planning system (TPS). This QC tool functions based on both PINNACLE scripting elements and PERL sub-routines. The core of this technique is the method of dynamic scripting, which involves a PERL programming module that is flexible and powerful for treatment plan data handling. Run-time plan data are collected, saved into temporary files, and analyzed against standard values and predefined logical rules. The results were summarized in a hypertext markup language (HTML) report that is displayed to the user. RESULTS This tool has been in clinical use for over a year. The occurrence frequency of technical problems, which would cause delays and suboptimal plans, has been reduced since clinical implementation. CONCLUSIONS In addition to drastically reducing the set of human-driven logical comparisons, this QC tool also accomplished some tasks that are otherwise either quite laborious or impractical for humans to verify, e.g., identifying conflicts amongst IMRT optimization objectives.

Quantitative metrics for assessing plan quality.

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

Vision 20/20: Automation and advanced computing in clinical radiation oncology

This Vision 20/20 paper considers what computational advances are likely to be implemented in clinical radiation oncology in the coming years and how the adoption of these changes might alter the practice of radiotherapy. Four main areas of likely advancement are explored: cloud computing, aggregate data analyses, parallel computation, and automation. As these developments promise both new opportunities and new risks to clinicians and patients alike, the potential benefits are weighed against the hazards associated with each advance, with special considerations regarding patient safety under new computational platforms and methodologies. While the concerns of patient safety are legitimate, the authors contend that progress toward next-generation clinical informatics systems will bring about extremely valuable developments in quality improvement initiatives, clinical efficiency, outcomes analyses, data sharing, and adaptive radiotherapy.

Knowledge-based prediction of three-dimensional dose distributions for external beam radiotherapy

PURPOSE To demonstrate knowledge-based 3D dose prediction for external beam radiotherapy. METHODS Using previously treated plans as training data, an artificial neural network (ANN) was trained to predict a dose matrix based on patient-specific geometric and planning parameters, such as the closest distance (r) to planning target volume (PTV) and organ-at-risks (OARs). Twenty-three prostate and 43 stereotactic radiosurgery/radiotherapy (SRS/SRT) cases with at least one nearby OAR were studied. All were planned with volumetric-modulated arc therapy to prescription doses of 81 Gy for prostate and 12-30 Gy for SRS. Using these clinically approved plans, ANNs were trained to predict dose matrix and the predictive accuracy was evaluated using the dose difference between the clinical plan and prediction, δD = Dclin - Dpred. The mean (〈δDr〉), standard deviation (σδDr ), and their interquartile range (IQR) for the training plans were evaluated at a 2-3 mm interval from the PTV boundary (rPTV) to assess prediction bias and precision. Initially, unfiltered models which were trained using all plans in the cohorts were created for each treatment site. The models predict approximately the average quality of OAR sparing. Emphasizing a subset of plans that exhibited superior to the average OAR sparing during training, refined models were created to predict high-quality rectum sparing for prostate and brainstem sparing for SRS. Using the refined model, potentially suboptimal plans were identified where the model predicted further sparing of the OARs was achievable. Replans were performed to test if the OAR sparing could be improved as predicted by the model. RESULTS The refined models demonstrated highly accurate dose distribution prediction. For prostate cases, the average prediction bias for all voxels irrespective of organ delineation ranged from -1% to 0% with maximum IQR of 3% over rPTV ∈ [ - 6, 30] mm. The average prediction error was less than 10% for the same rPTV range. For SRS cases, the average prediction bias ranged from -0.7% to 1.5% with maximum IQR of 5% over rPTV ∈ [ - 4, 32] mm. The average prediction error was less than 8%. Four potentially suboptimal plans were identified for each site and subsequent replanning demonstrated improved sparing of rectum and brainstem. CONCLUSIONS The study demonstrates highly accurate knowledge-based 3D dose predictions for radiotherapy plans.

Radiation sparing of cerebral cortex in brain tumor patients using quantitative neuroimaging

BACKGROUND AND PURPOSE Neurocognitive decline in brain tumor patients treated with radiotherapy (RT) may be linked to cortical atrophy. We developed models to determine radiation treatment-planning objectives for cortex, which were tested on a sample population to identify the dosimetric cost of cortical sparing. MATERIAL AND METHODS The relationship between the probability of cortical atrophy in fifteen high-grade glioma patients at 1-year post-RT and radiation dose was fit using logistic mixed effects modeling. Cortical sparing was implemented using two strategies: region-specific sparing using model parameters, and non-specific sparing of all normal brain tissue. RESULTS A dose threshold of 28.6 Gy was found to result in a 20% probability of severe atrophy. Average cortical sparing at 30 Gy was greater for region-specific dose avoidance (4.6%) compared to non-specific (3.6%). Cortical sparing resulted in an increase in heterogeneity index of the planning target volume (PTV) with an average increase of 1.9% (region-specific) and 0.9% (non-specific). CONCLUSIONS We found RT doses above 28.6 Gy resulted in a greater than 20% probability of cortical atrophy. Cortical sparing can be achieved using region-specific or non-specific dose avoidance strategies at the cost of an increase in the dose heterogeneity of the PTV.

Overview of the American Society for Radiation Oncology–National Institutes of Health–American Association of Physicists in Medicine Workshop 2015: Exploring Opportunities for Radiation Oncology in the Era of Big Data

Big data research refers to the collection and analysis of large sets of data elements and interrelationships that are difficult to process with traditional methods. It can be considered a subspecialty of the medical informatics domain under data science and analytics. This approach has been used in many areas of medicine to address topics such as clinical care and quality assessment (1–3). The need for informatics research in radiation oncology emerged as an important initiative during the 2013 National Institutes of Health (NIH)–National Cancer Institute (NCI), American Society for Radiation Oncology (ASTRO), and American Association of Physicists in Medicine (AAPM) workshop on the topic “Technology for Innovation in Radiation Oncology” (4). Our existing clinical practice generates discrete, quantitative, and structured patient-specific data (eg, images, doses, and volumes) that position us well to exploit and participate in big data initiatives. The well-established electronic infrastructure within radiation oncology should facilitate the retrieval and aggregation of much of the needed data. With additional efforts to integrate structured data collection of patient outcomes and assessments into the clinical workflow, the field of radiation oncology has a tremendous opportunity to generate large, comprehensive patient-specific data sets (5). However, there are major challenges to realizing this goal. For example, existing data are presently housed across different platforms at multiple institutions and are often not stored in a standardized manner or with common terminologies to enable pooling of data. In addition, many important data elements are not routinely discretely captured in clinical practice. There are cultural, structural, and logistical challenges (eg, computer compatibility and workflow demands) that will make the dream of big data research difficult. The big data research workshop provided a forum for leaders in cancer registries, incident report quality-assurance systems, radiogenomics, ontology of oncology, and a wide range of ongoing big data and cloud computing development projects to interact with peers in radiation oncology to develop strategies to harness data for research, quality assessment, and clinical care. The workshop provided a platform to discuss items such as data capture, data infrastructure, and protection of patient confidentiality and to improve awareness of the wide-ranging opportunities in radiation oncology, as well as to enhance the potential for research and collaboration opportunities with NIH on big data initiatives. The goals of the workshop were as follows: To discuss current and future sources of big data for use in radiation oncology research, To identify ways to improve our current data collection methods by adopting new strategies used in fields outside of radiation oncology, and To consider what new knowledge and solutions big data research can provide for clinical decision support for personalized medicine.

Automated radiation therapy treatment plan workflow using a commercial application programming interface

PURPOSE The objective of this study was to create a workflow for the automation and standardization of treatment plan generation and evaluation using an application programming interface (API) to access data from a commercial treatment planning system (Varian Medical Systems, Inc, Palo Alto, CA). METHODS AND MATERIALS The automation workflow begins with converting electronic patient-specific physician treatment planning orders that specify demographics, simulation instructions, and dosimetric objectives for targets and organs at risk into XML files. These XML files are used to generate standard contour names, beam, and patient-specific intensity modulated radiation therapy (IMRT) optimization templates to be executed in a commercial treatment planning system (TPS) by the user. A set of computer programs have been developed to provide quality control (QC) reports that verify demographic information in the TPS against the treatment planning orders, ensure the existence and proper naming of organs at risk, and generate patient-specific plan evaluation reports that provide real-time feedback on the concordance of an active treatment plan to the physician-specified treatment planning goals. RESULTS A workflow for lung IMRT was chosen as a test scenario. Contour, beam, and patient-specific IMRT optimization templates were automatically generated from the physician treatment planning orders and loaded into the planning system. The QC reports were developed for lung IMRT, including the option of patient-specific modifications to the standard templates. The API QC reporting includes a dynamic program that runs in parallel to the TPS during the planning process, providing real-time feedback as to whether physician-specified treatment plan parameters have improved or worsened from previous iterations. CONCLUSIONS User-created computer programs to access information in the TPS database by means of a commercial TPS API enable automation and standardization of treatment plan generation and evaluation.