Evaluation of the Dosimetric Accuracy of Brain Stereotactic Radiotherapy by Using a Hybrid Quality Assurance (QA) Toolkit

We evaluated the feasibility of a magnetic resonance (MR) tumor model and the accuracy of 3D dosimetry for evaluating brain stereotactic radiotherapy (SRT). We first validated the stability of the BANG3 polymer gel dosimeter and performed a pre-clinical verification of this device for brain SRT. Gel-filled vials were irradiated with 6-MV beams to generate a dose calibration curve in relation to the R2 (1/T2) values on 9.4 T MR images. The dose linearity, inter-reproducibility and intra-reproducibility, and dose-rate dependence of the gel dosimeters were evaluated. We also developed an image tumor model for pre-clinical testing. A calibration kit for the gel dosimeter and two spherical tumor representations were fabricated by using 3D printing. Two planning target volumes (PTVs) were contoured onto the MR images of the tumor models comprised of dosimetric gel. To evaluate the geometric and dosimetric accuracy, we created a treatment plan such that the D95s for PTV1 and shifted PTV2 were more than the prescribed dose. The shifted PTV2 was produced by an intentional shift of 5 mm from the true target position. A 2 arc VMAT plan was created to deliver 35 Gy in 5 fractions. After irradiation, calibration vials and tumor model phantoms were scanned by using 9.4 T MRI, and the acquired images were then analyzed using ImageJ and DCMTK software libraries. Scanned MRI images of the tumor models were imported into a treatment planning system and registered to CT images. We also compared the agreement of results between the planned and the measured data in 1D (ion chamber), 2D (2D film), and 3D (gel dosimeter). The best dose linearity achieved was 0.99 (R2) at 180 TE (ms) under which conditions the reproducibility and dose rate dependence were less than 2.2% and 3.5%, respectively. The mean dose differences between the treatment plan and the ion-chamber results were 3.3%, 0.4%, and 4.5%, respectively, for PTV1 and 2 and for the shifted PTV. Moreover, the point dose differences between the treatment plan and the gel results were 0.9%, 2.6% and 3.7%, respectively, for PTV1 and 2 and for the shifted PTV. The gamma passing rates with 3%/3 mm criteria were greater than 99% for all plans. The isodose distributions and profiles showed good qualitative agreement between the gel dosimeter, EBT film and RTP data for all PTVs. Our present study results, thus, indicate that an MR tumor model can effectively evaluate the geometric and dosimetric accuracy of brain SRT. Image tumor models using gel dosimeters are, therefore, useful for validating the 3D dose distributions for patient-specific whole treatment processes. We are planning to improve this image tumor model by facilitating respiratory mobility and utilizing more complex tumor shapes for liver stereotactic body radiotherapy (SBRT) cases.