No abstract

Purpose:The accuracy of dose prediction is essential for knowledge-based planning and automated planning techniques. We compare the dose prediction accuracy of 3 prediction methods including statistical voxel dose learning, spectral regression, and support vector regression based on limited patient training data.Methods:Statistical voxel dose learning, spectral regression, and support vector regression were used to predict the dose of noncoplanar intensity-modulated radiation therapy (4π) and volumetric-modulated arc therapy head and neck, 4π lung, and volumetric-modulated arc therapy prostate plans. Twenty cases of each site were used for k-fold cross-validation, with k = 4. Statistical voxel dose learning bins voxels according to their Euclidean distance to the planning target volume and uses the median to predict the dose of new voxels. Distance to the planning target volume, polynomial combinations of the distance components, planning target volume, and organ at risk volume were used as features for spectral regression and support vector regression. A total of 28 features were included. Principal component analysis was performed on the input features to test the effect of dimension reduction. For the coplanar volumetric-modulated arc therapy plans, separate models were trained for voxels within the same axial slice as planning target volume voxels and voxels outside the primary beam. The effect of training separate models for each organ at risk compared to all voxels collectively was also tested. The mean squared error was calculated to evaluate the voxel dose prediction accuracy.Results:Statistical voxel dose learning using separate models for each organ at risk had the lowest root mean squared error for all sites and modalities: 3.91 Gy (head and neck 4π), 3.21 Gy (head and neck volumetric-modulated arc therapy), 2.49 Gy (lung 4π), and 2.35 Gy (prostate volumetric-modulated arc therapy). Compared to using the original features, principal component analysis reduced the 4π prediction error for head and neck spectral regression (−43.9%) and support vector regression (−42.8%) and lung support vector regression (−24.4%) predictions. Principal component analysis was more effective in using all/most of the possible principal components. Separate organ at risk models were more accurate than training on all organ at risk voxels in all cases.Conclusion:Compared with more sophisticated parametric machine learning methods with dimension reduction, statistical voxel dose learning is more robust to patient variability and provides the most accurate dose prediction method.

Purpose: The Elements Spine Stereotactic Radiosurgery treatment planning system uses automated volumetric modulated arc radiotherapy that can provide a highly conformal dose distribution to targets, which can provide superior sparing of the spinal cord. This study compares the dosimetric quality of Elements plans with the clinical plans of 20 spine stereotactic radiosurgery/stereotactic body radiation therapy (SRS/SBRT) patients treated at our institution. Methods: Twenty spine SRS/SBRT patients who were clinically treated at our institution were replanned using the automated Elements planning workflow with prespecified templates. Elements automatically evaluates the size and shape of the target to determine if splitting the PTV into simplistic subvolumes, each treated by their own arc(s), would increase conformity and spinal cord sparing. The conformity index, gradient index, PTV D 5% , and maximum and mean cord dose were evaluated for the Elements and clinical plans. Treatment delivery efficiency was also analyzed by comparing the total number of monitor units and the modulation factor. Wilcoxon rank-sum tests were performed on the statistics. Results: Elements split the PTV for 50% of cases, requiring four or six arcs. Overall, Elements plans were found to be superior to clinical plans in conformity index, gradient index, and maximum cord dose. The PTV D 5% and cord mean dose for the Elements plans trended higher and lower, respectively. The numbers of monitor units and modulation factor were also higher for Elements plans, although the differences were not significant. Conclusion: Automated Elements plans achieved superior conformity and cord dose sparing compared to clinical plans and PTV splitting successfully improved spinal cord sparing.

Purpose Non‐coplanar 4π radiotherapy generalizes intensity modulated radiation therapy (IMRT) to automate beam geometry selection but requires complicated hyperparameter tuning to attain superior plan quality, which can be tedious and inconsistent. In this study, a fully automated 4π treatment planning was developed using evolving knowledge‐base (EKB) planning guided by dose prediction. Methods Twenty 4π lung and twenty 4π head and neck (HN) cases were included. A statistical voxel dose learning model was initially trained on low‐quality plans created using generic hyperparameter templates without manual tuning. To improve the automated plan quality without being limited by the training data quality, a new 4π optimization problem was formulated to include a one‐sided penalty on the organ‐at‐risk (OAR) dose deviation from the predicted dose. This directional OAR penalty encourages superior OAR sparing. The fast iterative shrinkage‐thresholding algorithm (FISTA) was used to solve the large‐scale beam orientation optimization problem. With the improved plans, new predictions were created to guide the next loop of EKB planning for a total of 10 loops. Plan quality was evaluated using a plan quality metric (PQM) points system based on clinical dose constraints and compared with automated planning approaches guided by manual high‐quality plans using all non‐coplanar beams, automated plans using individually evolved targeted dose, and manually created 4π plans. Results For the lung cases, the final EKB plans had significantly higher PQM than manually created 4π (+2.60%). The improvements plateaued after the third loop. The final HN EKB plans and manually created 4π plans had comparable PQMs, but had lower PQM compared to automated plans using a high‐quality training set (−3.00% and −4.44%, respectively). The PQM consistently increased up to the sixth loop. Individually evolved plans were able to improve the plan quality from initial condition due to the one‐sided cost function but the 60% of them were trapped in undesired local minima that were substantially worse than their corresponding EKB plans. Conclusion Evolving knowledge‐base planning is a novel automated planning technique guided by the predicted three‐dimensional dose distribution, which can evolve from low‐quality plans. EKB allows new beams to be used in the automated planning workflow for superior plan quality.

PurposeThe thermoluminescence dosimeter (TLD) has desirable features including low cost, reusability, small size, and relatively low energy dependence. However, the commonly available poly‐crystal TLDs (e.g., TLD‐100) exhibit high interdetector variability that requires individual calibration for high detection accuracy. To improve individual TLD tracking robustness, we developed an optical fingerprinting method to identify the TLD‐100 chips.MethodsSeven hundred and fifty‐two images were initially captured using a digital microscope camera to build a feature library for both facets of 376 TLD‐100 chips. A median intensity thresholding method was used to segment images into foreground and background. The affine transformation was used to register the segmented images to the same position. The fingerprint of each image was calculated from its registered image. All fingerprints were then recorded in an Elasticsearch® search database. The TLD fingerprint match was tested three times when the library was established and repeated once 20 months later. All chips were irradiated at 0, 1, 4, and 8 Gy on a calibrated clinical MV linac to establish the individual calibration curve.ResultsThe true positive rate of identifying TLDs based on their optical fingerprints was 100% at initialization of the inventory. After 20 months and multiple deployments for characterization, calibration, and dose measurement, the true positive match rate dropped to 99% with zero false‐positive matches. The TLDs exhibited high self‐consistency in the dose–response test with R2 between 0.988 and 1 with linear regression.ConclusionsThe TLD‐100 chips surface textures are unique and sufficient to support accurate identification based on the optical fingerprinting. This method provides inexpensive and robust management of the TLDs for individual calibration and dosimetry.