Purpose: To verify that optimized dose distributions provided by an intensity-modulated radiation therapy (IMRT) system are delivered accurately to human patients.
Methods and materials: Anthropomorphic phantoms are used to measure IMRT doses. Four types of verification are developed for: I) system commissioning with beams optimized to irradiate simulated targets in phantoms, II) plans with patient-optimized beams directed to phantoms simulating the patient, III) patient-phantom hybrid plans with patient-optimized beams calculated in phantom without further optimization, and IV) in vivo measurements. Phantoms containing dosimeters are irradiated with patient-optimized beams. Films are scanned and data were analyzed with software. Percent difference between verified and planned maximum target doses is defined as "dose discrepancy" (deltavp). The frequency distribution of type II deltavp from 204 verification films of 92 IMRT patients is fit to a Gaussian. Measurements made in vivo yield discrepancies specified as deltaivp, also fit to a Gaussian.
Results and discussion: Verification methods revealed three systematic errors in plans that were corrected prior to treatment. Values of [deltavp] for verification type I are <2%. Type II verification discrepancies are characterized by a Gaussian fit with a peak 0.2% from the centroid, and 158 [deltavp] <5%. The 46 values of [deltavp] >5% arise from differences between phantom and patient geometry, and from simulation, calculation, and other errors. Values of [deltavp] for verification III are less than half of the values of [deltavp] for verification II. A Gaussian fit of deltaivp from verification IV shows more discrepancy than the fit of deltavp, attributed to dose gradients in detectors, and exacerbated by immobilization uncertainty.
Conclusions: Dosimetric verification is a critical step in the quality assurance (QA) of IMRT. Hybrid Verification III is suggested as a preliminary quality standard for IMRT.