Utilizing Knowledge from Prior Plans in the Evaluation of Quality Assurance and Physics Considerations for Single-Isocenter Volumetric Modulated Arc Radiosurgery for Treatment of Multiple Intracranial Targets
Purpose: (Project 1): Increased interest in the Radiation Oncology Physics community regarding sensitivity of pre-treatment IMRT/VMAT QA to delivery errors has led to the development of DVH based analysis for pre-treatment QA. This paradigm shift necessitates a change in the acceptance criteria and action tolerance for QA. Here we present a knowledge based technique to objectively compare degradations in the DVH to the range from prior clinically accepted plans after adapting to the new patient's anatomy. We apply this knowledge based method to low risk prostate radiotherapy.
(Project 2): To address many of the physics challenges associated with single isocenter radiosurgery for multiple intracranial metastases (SIRMIT). Because the Varian High Definition MLC has variable leaf width with thicker leaves located >4cm from the isocenter (0.5cm vs. 0.25cm), this raises the question of whether the dose falloff and plan quality is inferior for targets located distal from the isocenter (>4cm). We hypothesize that such an effect will be greater for smaller targets, and we test this hypothesis and evaluate various isocenter placement strategies including one that favors placement of smaller targets closer to the isocenter.
Methods and Materials: (Project 1): DVHs for relevant organs at risk from a population of prior patients' plans were adapted using a machine learning algorithm to establish the DVH range specific to the patient's anatomy. The population of prior plans consisted of 198 (for OARs) and 40 (for PTVs) prostate cancer patients that had previously been planned and treated using IMRT. We then applied this technique to evaluate for single arc VMAT plans the clinical effect of six types of delivery errors: systematic offsets in a single centrally located MLC leaf; systematic leaf bank offsets, random normally distributed fluctuations in MLC position, systematic lag in gantry angle of the MLCs behind their intended position(s), fluctuations in dose rate, and delivery of each VMAT arc with a constant rather than a variable dose rate.
(Project 2): 11 SIRMIT plans with the isocenter placed at the PTV centroid were retrospectively analyzed to determine the relationship between relevant dosimetric indices and distance from the isocenter (e.g. relative tumor volume). We investigated three isocenter placement strategies for four prior SIRMIT patients that had larger variations in target volume; these strategies include (1) centroid, (2) Eclipse's built-in method, and (3) an "inverse center of mass" (ICM) method that weights the isocenter placement more heavily towards smaller lesions. Three VMAT SIRMIT plans were prepared with and without avoiding entrance / exit geometry through the eyes. Dose was calculate within the TPS and measured on an anthropomorphic phantom using Optically Stimulated Luminescence (OSL) dosimeters.
Results: (Project 1): QUANTEC suggests V75Gy dose limits of 15% for the rectum and 25% for the bladder, however the knowledge based constraints were more stringent: 8.48±2.65% for the rectum and 4.90±1.98% for the bladder. 19±10mm single leaf and 1.9±0.7mm single bank offsets resulted in rectum DVHs worse than 97.7% (2σ) of clinically accepted plans; all other errors fell within the clinically acceptable (2σ) rectum range. PTV degradations fell outside of the acceptable range for 0.6±0.3mm leaf offsets, 0.11±0.06mm bank offsets, 0.6±1.3mm of random noise, and 1.0±0.7° of gantry-MLC lag.
(Project 2): Values of conformity and gradient fall-off tended to be of higher quality for larger tumors (>1cc) close to isocenter. Moving the isocenter towards smaller tumors (ICM isocenter placement) yielded increased conformity and decreased gradient indices for these metastases. However, we observed a greater decrease in conformity and increase in gradient indices for larger lesions. Considering arc geometry entrance and avoidance angles for the eyes reduces lens dose from around 0.5-2.3Gy to ≤0.1Gy for a 20Gy SIRMIT plan.
Conclusions: (Project 1): Utilizing a group of prior treatment plans, a machine learning algorithm may be used to determine the distribution for each DVH that would have been achieved had the prior plans been prepared using the new patient anatomy, and degradations leading to statistical outliers may be identified. Using a knowledge based approach to QA evaluation enables customized QA criteria per treatment site, institution and or physician; and can often be more sensitive to errors than criteria based on organ complication rates.
(Project 2): CI and GI values were poorer for small distal targets, especially >6cm. Large tumors benefited from centroid placement more than small tumors did from ICM. It is necessary to consider arc geometry avoidance angles to adequately reduce lens dose.
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Masters Theses