Design and Construction of Heterogeneity Phantom for Comparison of Dose Calculation Algorithms of Treatment Planning Systems

ABSTRACT
 Verification of the output of a treatment planning system (TPS) is considered to be part of quality assurance (QA) requirement for the TPS and need to be performed during commissioning and annually, but this is not given the needed priority due to non availability of appropriate phantoms to facilitate the process. Phantoms with heterogeneities that can simulate clinical situations are needed to assess the performance of TPSs. This research work sought to design and construct such a phantom from locally available materials to be used as QA/quality control (QC) tool for radiotherapy treatment planning systems. The constructed phantom could be configured to represent tissue-lungtissue or tissue-bone-tissue anatomical regions. The phantom was composed of Polymethylmethacrylate (PMMA) slabs, slabs of wood and a Perspex compartment completely filled with Portland cement, which were used to mimic tissue, lung and bone,
respectively. The Perspex slabs, the slabs of wood and the Perspex compartment had the following dimensions (length   width    thickness): 30 cm   30 cm   1 cm, 30 cm   30 cm   2 cm and 30 cm   30 cm   5 cm, respectively. A Perspex slab with dimensions (length   width   thickness): 30 cm   30 cm   2 cm with a hole to accommodate a 0.125 cc PTW Semiflex ionization chamber on one of its sides was also constructed to be part of the phantom. One of the slabs of wood also had a hole on one of its sides similar to that of the Perspex slab. The phantom was formed by stacking the various constituent components together to give the desired configuration. After construction, the phantom was used to assess the performance of three TPSs using different dose calculation algorithms. The phantom was scanned in three different configurations with Toshiba Aquilion One CT-scanner and the images were loaded onto a CD-ROM. The images were later transferred to a Prowess Panther (version 4.6), Oncentra MasterPlan (version 4.3), and Varian Eclipse (version 13.6) TPSs via available DICOM import tools. Treatment plans were created for the different configurations of the phantom per TPS using the treatment machine in use with the TPS for patient treatment planning. The same treatment plans were created for each of the TPSs.  For each TPS, dose distributions were calculated with all available dose calculation algorithms. The TPSs considered used Convolution-Superposition (CS), Collapsed Cone Convolution (CCC), Pencil Beam (PB) and the Analytical Anisotropic Algorithm (AAA) algorithms. The created treatment plans were replicated on the respective treatment machines and doses at calculation points placed within the phantom during the treatment planning processes measured with a 0.125 cc PTW Semiflex ionization chamber. The measured doses were compared to their
calculated counterparts by the TPSs, and the differences in the doses expressed as a percentage of the respective measured doses.  
The results of the dose comparison showed that for lung region, the percentage difference between the measured and the calculated doses ranged from: -0.27 to 4.64% (mean of 1.87±1.72%), -10.22 to 9.24% (mean of 4.33±3.81%), -9.13 to 8.39% (mean of 5.83±2.94%) and -3.83 to 9.28% (mean of 3.09±2.83%) for doses calculated with AAA, CCC, PB, and CS, respectively. And for the bony region, the percentage difference between the measured and the calculated doses ranged from: -1.52 to 4.64% (mean of 1.83±1.52%), -13.61 to 4.39% (mean of 5.85±5.20%), -14.52 to 4.81% (mean of 5.97±5.63%) and -10.55 to -3.48% (mean of 5.91±3.08%) for doses calculated with AAA, CCC, PB, and CS, respectively. For dose calculation points placed within high dose gradient regions, the dose calculation algorithms gave discrepancies in dose greaterthan 80% with the exception of AAA, which gave percentage difference in dose close to 20%. The constructed phantom had provided a cost effective way of accessing the outputs of dose calculation algorithms of TPSs. Using the phantom it was shown that AAA was the most accurate dose calculation algorithm among those considered in the study. The use of the constructed phantom for clinical practice is recommended.