An evaluation of patient-specific IMRT verification failures

• Main Author: Crawford, Jason
• Other Authors: Gagne, Isabelle Marie
• Language: English; en
• Published: 2010
• Subjects: IMRT; EPID; verification; Portal image dose reconstruction; Monte Carlo; Fluence modulation; intensity modulated radiation therapy; ion chamber; UVic Subject Index::Sciences and Engineering::Physics::Radiation; UVic Subject Index::Sciences and Engineering::Physics; UVic Subject Index::Sciences and Engineering::Health Sciences::Oncology
• Online Access: http://hdl.handle.net/1828/3041

At the BC Cancer Agency (BCCA), Vancouver Island Centre (VIC), the clinical verification of Intensity Modulated Radiation Therapy (IMRT) treatment plans involves comparing Portal Image (PI) -based three-dimensionally reconstructed (EPIDose) dose distributions to planned doses calculated using the Pencil Beam Convolution (PBC) algorithm. Discrepancies surpassing established action levels constitute failure. Since 2007, the failure rate of IMRT verification process had been increasing, reaching as high as 18.5% in 2009. A retrospective evaluation of clinical IMRT verification failures was conducted to identify causes and possible resolutions. Thirty clinical verification failures were identified. An equipment malfunction was discovered and subsequently repaired, and several failures were resolved in the process. Statistical uncertainty in measurement outcome was small in comparison to action levels and not considered significant to the production of failures. Still, over 50% of the redelivered plans were shown to consistently fail. A subgroup of consistent verification plans were compared to ion chamber point dose measurements. Relative to ion chamber measurements, EPIDose underestimated the dose while the dose calculation algorithm (PBC, Eclipse version 8.1.18) overestimated the same point dose. Comparisons of individual fields demonstrated that none were identifiably problematic; dose discrepancies were the result of minor but accumulating dose differences. Consistent verification failures were recalculated using two advanced dose calculation engines (the Anisotropic Analytical Algorithm and Monte Carlo). In general, verification metrics improved, and all failures were resolved. Three distinct indices of fluence modulation (or complexity) were shown to correlate with verification metrics. This indicated that deficiencies in both the leaf motion calculator and the PBC (Eclipse version 8.1.18) had likely contributed to the production of failures. In conclusion, clinical verification failures were resolved retrospectively by replacing faulty equipment and using more advanced methods of planned dose calculation, supporting the efficacy and continued use of PI-based three dimensional dose reconstruction for IMRT verification.