Purpose: To show the benefits of deriving meaningful tolerance tables from data, and how the proper number of warnings could improve setup consistency.
Methods: One year of historical patient positions were collected from the record and verify database. A set of new tolerance tables for different anatomical sites was created based on twice the standard deviation of the corresponding patient positions. In most scenarios, the new tolerance tables were more relaxed than the historical ones, and they were subsequently implemented for clinical use. To investigate the impacts of the new tolerance tables, three months of patient positions controlled by either the historical or new tolerance tables were studied and compared.
Results: For photon beam therapy, the couch override frequencies for patient positions were 6.7% lateral, 11.9% longitudinal, and 2.8% vertical using the historical tolerance tables. This was compared to 2.1% lateral, 1.8% longitudinal, and 0.4% vertical for the new tables. The standard deviations in patient positions were 1.13 cm lateral, 2.43 cm longitudinal, and 0.29 cm vertical for the former, versus 1.07 cm lateral, 2.04 cm longitudinal, and 0.29 cm vertical for the latter. In general, using the new tolerance tables resulted in smaller lateral and longitudinal standard deviations (p < 0.01). As for electron therapy, the override frequencies were 4.4% lateral, 8.5% longitudinal, and 6.1% vertical using the historical tolerance tables, and 5.9% lateral, 3.0% longitudinal, and 2.6% vertical using the new tables. The standard deviations in patient positions were 1.75 cm lateral, 4.00 cm longitudinal, and 0.74 cm vertical for the former, versus 1.17 cm lateral, 2.23 cm longitudinal, and 0.51 cm vertical for the latter. The differences in all three dimensions were significant (p < 0.01).
Conclusion: Applying statistically derived tolerance tables, even with more relaxed tolerances, can significantly reduce the variation in patient positions.