1. A novel approach to EPID-based 3D volumetric dosimetry for IMRT and VMAT QA
Alhazmi et al. developed a novel EPID-based approach to reconstruct a 3D dose distribution for patient-specific pre-treatment dosimetric QA of IMRT and VMAT plans. The 2D dose distributions were acquired in air and then converted to 2D dose distributions in water for each depth. The integrated 3D dose distribution was obtained by summing the reconstructed 3D dose distributions during treatment delivery. The reconstructed 3D dose distributions of the treatment plans were compared against the dose distributions measured from Octavius phantom using 3%/3mm gamma criteria with results showing 99% and 96% pass rate for static step-and-shoot IMRT and VMAT respectively.
Phys. Med. Biol. May 2018, 63 (11), pp: 1 – 13

2. A fast optimization approach for treatment planning of volumetric modulated arc therapy
Yan et al. proposed a novel, fast optimization algorithm for VMAT planning. The algorithm utilizes a progressive sampling strategy, starting with the coarse sampling first and then increasing the sampling resolution continuously. Authors used a gradient-based fluence-map optimization followed by a mixed integer linear programming algorithm for leaf sequencing. The new VMAT optimization algorithm was evaluated using a head and neck, a lung and a prostate. IMRT cases planned with a conventional IMRT method using a commercial TPS. For all cases, the quality of the VMAT plans are similar or better than the ones obtained with the IMRT plans. As expected, much better sparing of the OARs is achieved with the VMAT. In addition, the treatment delivery time of <5 min was achieved with the VMAT.
Radiation Oncology. May 2018, 13:101

3. The impact of robustness of deformable image registration on contour propagation and dose accumulation for head and neck adaptive radiotherapy
Zhang et al. evaluated the accuracy of two commercial DIR algorithms in Ray Station TPS as well as their ability to propagate contours and accumulate dose in Head and Neck ART. In this retrospective study, authors selected ten head and neck cancer patients who had weekly CT scans and received offline ART during the course of radiotherapy. The deformed and radiation oncologist-drawn contours on weekly CT scans were compared. The weekly doses were accumulated back to planning CT based on the two DIR algorithms. The results showed Dice > 0.8 and HD < 0.1cm and mean dose variation < 60 cGy using the detailed settings of both DIR algorithms. However, simple settings in both DIR algorithms resulted in large geometric and dosimetric variations. Overall, this study showed that there were significant differences in both geometric and dosimetric parameters between two DIR algorithms and that individual parameter settings in each DIR algorithm found to have a significant impact on the results of the DIRs.
J Appl Clin Med Phys. May 2018, Volume 101, Issue 2, Pages 1-10.

4. Utilizing skin sparing technique in HN VMAT treatment
Spencer et al. investigated the reduction in skin dose using a skin sparing technique for ten head and neck patients treated with VMAT. The technique utilized a 5mm thick skin sparing structure which was created by subtracting wall from the 5Gy isodose line structure. The VMAT plans were then reoptimized using the new skin sparing structure. The comparison of the VMAT plans which were optimized using skin sparing structure showed up to 13% reduction in skin doses as compared to the VMAT plans optimized without skin sparing structure. This study demonstrated that the use of a skin sparing structure in the VMAT optimization has potential to reduce the skin dose, therefore reducing the skin toxicity during the course of treatment.
Medical Dosimetry. May 2018, Volume 43.

5. Optimal collimator rotation based on the outline of multiple brain targets in VMAT
Kim et al. investigated the ideal collimator rotation which resulted in the best dosimetric plan quality and MU efficiency using twenty patients with multiple target regions. The technique utilizes the integrated MLC apertures which is based on the outline of multiple brain lesions. Study results showed that 400 collimator angle resulted in lowest point-averaged field size values which were 1.5 times smaller than the one obtained with the fixed collimator angle currently used with the conventional IMRT plans. Furthermore, the improved sparing of some OARs and reduction in MU were achieved for VMAT planning of multiple brain targets.
Radiation Oncology. May 2018, Volume 13:88 Pages 1-10.

6. Volumetric modulated arc therapy of head-and-neck cancer on a fast-rotating O-ring linac: Plan quality and delivery time comparison with a C-arm linac
Michiels et al. compared the plan quality achievable with new O-ring, jawless linac (Halcyon, Varian Medical Systems) to the conventional C-arm linac for 30 patients with head and neck cancer. While standard C-Arm linac VMAT plans utilized 6MV beam with 2 arcs the O-Ring VMAT plans used 6FFF beam with both 2 arcs and 3 arcs. The same optimization weights were used for both C-Arm and O-Ring based VMAT plans. Minor improvements in target dose homogeneity and OAR sparing were obtained using the O-Ring based 3 arc VMAT plans as compared to the 3 arc VMAT plans. The results also demonstrated that the O-Ring based linac maintains the dosimetric quality of the C-arm based 2 arc VMAT plans.
Radiotherapy and Oncology. May 2018

7. VMAT optimization with dynamic collimator rotation
Lyu et al. developed a VMAT optimization method for dynamically varying collimator angle during the arc motion. The so-called dynamically collimated VMAT (DC-VMAT) optimization alternates between direct aperture optimization (DAO) and collimator angle selection. Feasibility of DC-VMAT using one full-arc with dynamic collimator rotation was tested on a phantom with two small spherical targets, a brain, a lung and a prostate cancer patient. The plan was compared against a static collimator VMAT (SC-VMAT) plan using three full arcs with 60 degrees of collimator angle separation in patient studies. With the same target coverage, DC-VMAT achieved 20.3% reduction of R50 in the phantom study and reduced the average max and mean OAR dose by 4.49% and 2.53% of the prescription dose in patient studies, as compared with SC-VMAT. While DC-VMAT plans require slower gantry motion to accommodate multiple collimator angles, DC-VMAT with a single arc manages to achieve superior dosimetry than SC-VMAT with three full arcs.
Med. Phys. 45 (6), June 2018, pp.2399-2410

8. Interplay effect on a 6-MV flattening-filter-free linear accelerator with high dose rate and fast multi-leaf collimator motion treating breast and lung phantoms
Netherton et al. investigated the dosimetric delivery error (interplay effect) caused by the combined motions of patient anatomy and multileaf collimation (MLC) sequences of a new linear accelerator (Halcyon, Varian Medical Systems) having a high dose rate and fast MLC and gantry speeds. The authors developed IMRT and VMAT plans for breast and lung with varying complexity levels and arc numbers and measured delivered dose to MOSFET detectors implanted in motion phantoms under sinusoidal breathing motion simulations. The mean beam modulation for plans created for the Halcyon 1.0 linear accelerator was 2.9 MU/cGy (two- to four-field IMRT breast plans), 6.2 MU/cGy (at least five-field IMRT), and 3.6 MU/cGy (four-arc VMAT). Maximum and mean dose deviations increased with increasing plan complexity under tumor motion for breast and lung treatments. Concerning VMAT plans under motion, maximum, and mean dose deviations were higher for one arc than for two arcs regardless of plan complexity. To minimize dose deviations across multiple fractions for dynamic targets, the authors recommend limiting treatment plan complexity and avoiding one-arc VMAT on the Halcyon 1.0 linear accelerator when interplay effect is a concern.
Med. Phys. 45 (6), June 2018, pp.2369-2376

9. Increased accuracy of planning tools for optimization of dynamic multileaf collimator delivery of radiotherapy through reformulated objective functions
Engberg et al. investigated the potential of a new formulation of objective functions based on mean-tail dose and dynamic multileaf collimator (DMLC) deliverability constraints with an interior point method to efficiently solve the resulting multi-criteria optimization (MCO) to facilitate better automation in plan optimization than afforded by conventional planning objectives using penalty functions. The proposed formulation of objective functions and optimization strategy were tested on Pareto optimal DMLC plans for three patient cases: a prostate, a lung, and a head-and-neck case, subject to fulfilment of several PTV and OAR dose requirements expressed in terms of dose-at-volume, average, minimum and maximum dose limits. DMLC plans for the three clinical cases Pareto optimal to the proposed formulation compared well with those DMLC plans resultant from the penalty-based MCO module in RayStation, indicating the potential to streamline the planning process with appropriate objective functions, potentially eliminating the need for trial and error and time-consuming re-optimizations in the current planning process.
Phys. Med. Biol. 63 (2018) 125012 (12pp)

10. Comparative analysis for renal stereotactic body radiotherapy using Cyberknife, VMAT and proton therapy based treatment planning
Baydoun et al. carried out a retrospective analysis of stereotactic body Radiotherapy (SBRT) planning comparison for renal cell carcinoma (RCC) on five different platforms utilizing CyberKnife, VMAT, and pencil beam scanning (PBS) proton therapy at two institutions with an intention to evaluate the feasibility of multi-center RCC SBRT trial. Ten previously treated patients in a phase I trial were selected for this study. Using the anonymized CT data sets, treatment planning was performed across all five platforms with the goal of delivering 48 Gy to the PTV in five fractions. Treatment planning comparison produced comparable results for all platforms and techniques used. The study was limited to treatment planning aspects without looking into delivery and dose verification issues on different platforms and treatment units.
J Appl Clin Med Phys 2018; 19:3:125–130

11. Fully automated searching for the optimal VMAT jaw settings based on Eclipse Scripting Application Programming Interface (ESAPI) and RapidPlan knowledge-based planning
Huang et al. investigated automatically finding optimal VMAT jaw settings based on Eclipse Scripting Application Programming Interface (ESAPI) and RapidPlan knowledge-based planning in Eclipse workspace. Ten previously treated rectal cancer VMAT patient cases were selected to study the dosimetric impact of jaw settings on the VMAT planning and to find the best jaw configuration for MLC sequencing. The authors used ESAPI to create and evaluate a large number of plans automatically using various jaw settings such that X jaws were gradually extended toward the isocenter while the Y jaws were retracted by the width of an adjacent MLC leaf for adequate scatter contribution. The authors used a predeveloped and validated RapidPlan model to automate the assignment of objective functions and used a plan scoring function using a mean dose volume component and a high dose volume component corresponding to the mean dose and maximum dose to OARs, respectively. They confirmed the insensitivity of RapidPlan DVH estimate to actual jaw settings among plans with various jaws settings. Their scoring system showed that conformal jaw settings did not always produce the optimal VMAT planning.
J Appl Clin Med Phys 2018; 19:3:177–182

12. Comparison of MLC error sensitivity of various commercial devices for VMAT pre-treatment quality assurance
Saito et al. investigated sensitivity of detecting MLC delivery errors by several commercial QA devices including Scandidos Delta4, PTW 2D-array, iRT systems IQM, and the PTW Farmer chamber. The authors retrospectively selected nine previously treated VMAT plans for SBRT, prostate, and head-and-neck sites and introduced systematic MLC leaf position errors ranging from -0.75 mm to 0.75 mm in steps of 0.25 mm as well as a random error of 0.5 mm into MLC delivery files for selected VMAT plans. They evaluated the sensitivity of each measurement device for detecting intentionally introduced errors with respect to baseline measurements from original plans without the errors and determined the global gamma passing rates at 1%/1, 2%/2, and 3%/3 mm and dose differences at 1%, 2%, and 3%. The authors found a strong linear correlation for the cumulative dose differences from the original plans measured by the two ion chambers as a function of introduced MLC errors, with a statistically significant higher sensitivity for the IQM than that of the farmer chamber. Both Delta4 and PTW 2D-array performed similarly in gamma analysis of pass rates. However, they were both found to be insensitive to small MLC positioning errors in the order of ±0.25 mm for small treatment fields such as SBRT. The authors also concluded that dose difference could be used as a better criterion than gamma for daily MLC QA.
J Appl Clin Med Phys 2018; 19:3:87–93

13. Creation of knowledge-based planning models intended for large scale distribution: Minimizing the effect of outlier plans
Aviles et al. proposed a process to properly identify outlier plans in creating knowledge-based planning (KBP) models and how to remove outliers from the model to increase accuracy and reliability of the model predictions for DVH calculations in prostate and head-neck test sites using the commercially available RapidPlan platform. They used statistical tools to classify outlier plans in three categories: geometric, dosimetric, and over-fitting outliers. They used an independent set of clinical plans than those used for training the model to validate DVH estimates obtained by the model. By removing the geometric and over-fitting outliers and replacing the dosimetric outliers after re-planning they were able to improve the accuracy of DVH estimation as measured by several DVH metrics (V50, V85, and V99) by -2% to 4% for the prostate model and -2.0% to 7.6% for the head-and-neck model.
J Appl Clin Med Phys 2018; 19:3:215–226