Encrypted login | home

Program Information

A Long-Term Study of Correlations Between Common Linac QC Test Results and Patient-Specific QC Measurements

D Granville

DA Granville*, D La Russa, J Belec, J Sutherland, The Ottawa Hospital Cancer Centre, Ottawa, ON


TH-AB-FS1-2 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: Four Seasons 1

Purpose: To quantify the correlations between common linac QC tests and patient-specific QC measurements for treatment plans of varying complexity.

Methods: We selected 4 VMAT treatment plans of varying complexity (eg varying dose, modulation, treatment volume, and off-axis distance) for this study. These plans represented treatments of the following sites: breast, pelvis, prostate, and lung SBRT. Dose distributions for each plan were measured approximately monthly using a biplanar diode array (ScandiDos Delta4) on 6 matched linacs (Elekta with Agility MLC) over a ~1 year period, for a total of 300 measurements (75 per plan). For each delivery of each plan, the median deviation between planned and measured doses was compared to the most recent linac QC results to assess their correlation. The following linac QC tests were included in the analysis: output, profile flatness, profile symmetry, and sliding window dose constancy. Correlation strengths were quantified using Pearson’s correlation coefficient.

Results: We found considerable differences in the relative correlation strength of each linac QC test with median dose deviation across the different treatment plans. Predictably, linac output had the strongest correlation for most of the treatment plans (0.58, 0.60, and 0.57 for pelvis, prostate and lung SBRT, respectively). Interestingly, the beam profile flatness had a stronger correlation with median dose deviation than output for the breast plan (0.68 vs 0.42), likely owing to the fact that this treatment is performed off-axis. Generally, for smaller centrally located treatment volumes, the correlation strength of linac output dominated that of the other QC tests. However, for larger treatment volumes extending further off the central axis, the correlation strength of the other tests increased relative to output.

Conclusion: The delivery accuracies of different VMAT treatment plans show markedly different dependence on variations in routine linac QC results.

Contact Email: