Deep learning-based dose prediction to improve the plan quality of volumetric modulated arc therapy for gynecologic cancers.

BACKGROUND: In recent years, deep-learning models have been used to predict entire three-dimensional dose distributions. However, the usability of dose predictions to improve plan quality should be further investigated. PURPOSE: To develop a deep-learning model to predict high-quality dose distributions for volumetric modulated arc therapy (VMAT) plans for patients with gynecologic cancer and to evaluate their usability in driving plan quality improvements. METHODS: A total of 79 VMAT plans for the female pelvis were used to train (47 plans), validate (16 plans), and test (16 plans) 3D dense dilated U-Net models to predict 3D dose distributions. The models received the normalized CT scan, dose prescription, and target and normal tissue contours as inputs. Three models were used to predict the dose distributions for plans in the test set. A radiation oncologist specializing in the treatment of gynecologic cancers scored the test set predictions using a 5-point scale (5, acceptable as-is; 4, prefer minor edits; 3, minor edits needed; 2, major edits needed; and 1, unacceptable). The clinical plans for which the dose predictions indicated that improvements could be made were reoptimized with constraints extracted from the predictions. RESULTS: The predicted dose distributions in the test set were of comparable quality to the clinical plans. The mean voxel-wise dose difference was -0.14 ± 0.46 Gy. The percentage dose differences in the predicted target metrics of D 1 % ${D}_{1{\mathrm{\% }}}$ and D 98 % ${D}_{98{\mathrm{\% }}}$ were -1.05% ± 0.59% and 0.21% ± 0.28%, respectively. The dose differences in the predicted organ at risk mean and maximum doses were -0.30 ± 1.66 Gy and -0.42 ± 2.07 Gy, respectively. A radiation oncologist deemed all of the predicted dose distributions clinically acceptable; 12 received a score of 5, and four received a score of 4. Replanning of flagged plans (five plans) showed that the original plans could be further optimized to give dose distributions close to the predicted dose distributions. CONCLUSIONS: Deep-learning dose prediction can be used to predict high-quality and clinically acceptable dose distributions for VMAT female pelvis plans, which can then be used to identify plans that can be improved with additional optimization.

Deep Learning-Based Dose Prediction for Automated, Individualized Quality Assurance of Head and Neck Radiation Therapy Plans.

PURPOSE: This study aimed to use deep learning-based dose prediction to assess head and neck (HN) plan quality and identify suboptimal plans. METHODS AND MATERIALS: A total of 245 volumetric modulated arc therapy HN plans were created using RapidPlan knowledge-based planning (KBP). A subset of 112 high-quality plans was selected under the supervision of an HN radiation oncologist. We trained a 3D Dense Dilated U-Net architecture to predict 3-dimensional dose distributions using 3-fold cross-validation on 90 plans. Model inputs included computed tomography images, target prescriptions, and contours for targets and organs at risk (OARs). The model's performance was assessed on the remaining 22 test plans. We then tested the application of the dose prediction model for automated review of plan quality. Dose distributions were predicted on 14 clinical plans. The predicted versus clinical OAR dose metrics were compared to flag OARs with suboptimal normal tissue sparing using a 2 Gy dose difference or 3% dose-volume threshold. OAR flags were compared with manual flags by 3 HN radiation oncologists. RESULTS: The predicted dose distributions were of comparable quality to the KBP plans. The differences between the predicted and KBP-planned D1%,D95%, and D99% across the targets were within -2.53% ± 1.34%, -0.42% ± 1.27%, and -0.12% ± 1.97%, respectively, and the OAR mean and maximum doses were within -0.33 ± 1.40 Gy and -0.96 ± 2.08 Gy, respectively. For the plan quality assessment study, radiation oncologists flagged 47 OARs for possible plan improvement. There was high interphysician variability; 83% of physician-flagged OARs were flagged by only one of 3 physicians. The comparative dose prediction model flagged 63 OARs, including 30 of 47 physician-flagged OARs. CONCLUSIONS: Deep learning can predict high-quality dose distributions, which can be used as comparative dose distributions for automated, individualized assessment of HN plan quality.

Automation of radiation treatment planning for rectal cancer.

PURPOSE: To develop an automated workflow for rectal cancer three-dimensional conformal radiotherapy (3DCRT) treatment planning that combines deep learning (DL) aperture predictions and forward-planning algorithms. METHODS: We designed an algorithm to automate the clinical workflow for 3DCRT planning with field aperture creations and field-in-field (FIF) planning. DL models (DeepLabV3+ architecture) were trained, validated, and tested on 555 patients to automatically generate aperture shapes for primary (posterior-anterior [PA] and opposed laterals) and boost fields. Network inputs were digitally reconstructed radiographs, gross tumor volume (GTV), and nodal GTV. A physician scored each aperture for 20 patients on a 5-point scale (>3 is acceptable). A planning algorithm was then developed to create a homogeneous dose using a combination of wedges and subfields. The algorithm iteratively identifies a hotspot volume, creates a subfield, calculates dose, and optimizes beam weight all without user intervention. The algorithm was tested on 20 patients using clinical apertures with varying wedge angles and definitions of hotspots, and the resulting plans were scored by a physician. The end-to-end workflow was tested and scored by a physician on another 39 patients. RESULTS: The predicted apertures had Dice scores of 0.95, 0.94, and 0.90 for PA, laterals, and boost fields, respectively. Overall, 100%, 95%, and 87.5% of the PA, laterals, and boost apertures were scored as clinically acceptable, respectively. At least one auto-plan was clinically acceptable for all patients. Wedged and non-wedged plans were clinically acceptable for 85% and 50% of patients, respectively. The hotspot dose percentage was reduced from 121% (σ = 14%) to 109% (σ = 5%) of prescription dose for all plans. The integrated end-to-end workflow of automatically generated apertures and optimized FIF planning gave clinically acceptable plans for 38/39 (97%) of patients. CONCLUSION: We have successfully automated the clinical workflow for generating radiotherapy plans for rectal cancer for our institution.

Using Failure Mode and Effects Analysis to Evaluate Risk in the Clinical Adoption of Automated Contouring and Treatment Planning Tools.

PURPOSE: In this study, we applied the failure mode and effects analysis (FMEA) approach to an automated radiation therapy contouring and treatment planning tool to assess, and subsequently limit, the risk of deploying automated tools. METHODS AND MATERIALS: Using an FMEA, we quantified the risks associated with the Radiation Planning Assistant (RPA), an automated contouring and treatment planning tool currently under development. A multidisciplinary team identified and scored each failure mode, using a combination of RPA plan data and experience for guidance. A 1-to-10 scale for severity, occurrence, and detectability of potential errors was used, following American Association of Physicists in Medicine Task Group 100 recommendations. High-risk failure modes were further explored to determine how the workflow could be improved to reduce the associated risk. RESULTS: Of 290 possible failure modes, we identified 126 errors that were unique to the RPA workflow, with a mean risk priority number (RPN) of 56.3 and a maximum RPN of 486. The top 10 failure modes were caused by automation bias, operator error, and software error. Twenty-one failure modes were above the action threshold of RPN = 125, leading to corrective actions. The workflow was modified to simplify the user interface and better training resources were developed, which highlight the importance of thorough review of the output of automated systems. After the changes, we rescored the high-risk errors, resulting in a final mean and maximum RPN of 33.7 and 288, respectively. CONCLUSIONS: We identified 126 errors specific to the automated workflow, most of which were caused by automation bias or operator error, which emphasized the need to simplify the user interface and ensure adequate user training. As a result of changes made to the software and the enhancement of training resources, the RPNs subsequently decreased, showing that FMEA is an effective way to assess and reduce risk associated with the deployment of automated planning tools.

Intensity Modulated Radiation Therapy Versus Volumetric Arc Radiation Therapy in the Treatment of Glioblastoma-Does Clinical Benefit Follow Dosimetric Advantage?

PURPOSE: Volumetric modulated arc therapy (VMAT) has been shown by multiple planning studies to hold dosimetric advantages over intensity modulated radiation therapy (IMRT) in the management of brain tumors, including glioblastoma (GBM). Although promising, the clinical impact of these findings has not been fully elucidated. METHODS AND MATERIALS: We retrospectively reviewed consecutive patients with a pathologic-confirmed diagnosis of GBM who were treated between 2014 and 2015, a period that encompassed the transition from IMRT to VMAT at a single institution. After surgery, radiation with VMAT consisted of 2 to 3 coplanar arcs with or without an additional noncoplanar arc or IMRT with 5 to 6 gantry angles with concurrent and adjuvant temozolomide. Actuarial analyses were performed using the Kaplan Meier method. RESULTS: A total of 88 patients treated with IMRT (n = 45) and VMAT (n = 43) were identified. Patients were similar in terms of age, sex, performance status, extent of resection, and the high dose target volume. At a median follow-up time of 27 months (range, .7-32.3 months), the overall survival, freedom from progression, and freedom from new or worsening toxicity rates were not different between the 2 treatment groups (log-rank: P = .33; .87; and .23, respectively). There was no difference in incidences of alopecia, erythema, nausea, worsening or new onset fatigue, or headache during radiation, or temozolomide dose reduction for thrombocytopenia or neutropenia (all P > .05). Patterns of failure were different with more out of field failures in the IMRT group (P = .02). The mean time of treatment (TOT) was significantly reduced by 29% (P < .01) with VMAT (mean TOT: 10.3 minutes) compared with IMRT (mean TOT: 14.6 minutes). CONCLUSIONS: For GBM, treatment with VMAT results in similar oncologic and toxicity outcomes compared with IMRT and may improve resource utilization by reducing TOT. VMAT should be considered a potential radiation modality for patients with GBM.

Incidence and predictors of chest wall toxicity after high-dose radiation therapy in 15 fractions.

PURPOSE: Fifteen fraction treatment schedules are increasingly used to deliver high doses of radiation therapy (RT) to both lung and hepatobiliary malignancies. The purpose of our study was to examine the incidence and predictors of chest wall (CW) toxicity in patients treated with this regimen. METHODS AND MATERIALS: We evaluated 135 patients treated with RT to doses ≥52.5 Gy in 15 fractions for thoracic and hepatobiliary malignancies between January 2009 and December 2012. We documented patient characteristics and CW dosimetric parameters for each case. Toxicity was scored using the Common Terminology Criteria for Adverse Events, version 4.0, criteria for radiation dermatitis and CW pain. Patient characteristics and CW dosimetric parameters were evaluated for their association with CW toxicity using proportional hazards regression. RESULTS: Median follow-up was 9 months from the start of RT. Forty-eight patients (36%) developed dermatitis at a median time of 18 days. In multivariable analysis, the absolute volume of CW (in cm3) receiving 40 Gy (V40) ≥120 cm3 was associated with the occurrence of dermatitis (hazard ratio, 3.12; 95% confidence interval, 1.74-5.60; P < .001). Twenty-one patients (16%) developed CW pain (20 grade 1, 1 grade 2) at a median time of 3 months. In multivariable analysis, CW V40 ≥150 cm3 was associated with the occurrence of CW pain (hazard ratio, 2.65; 95% confidence interval, 1.12-6.24; P = .03). The absolute rate of CW pain in patients with V40 <150 cm3 was 11% versus 26% in patients with V40 ≥150 cm3 (P = .03). CONCLUSIONS: Hypofractionated RT with 15 fraction regimens results in an acceptable incidence of CW toxicity, specifically CW pain. We recommend a dose constraint of V40 <150 cm3 to minimize this adverse event.