Site-specific tolerance tables and indexing device to improve patient setup reproducibility

While the implementation of tools such as image-guidance and immobilization devices have helped to prevent geometric misses in radiation therapy, many treatments remain prone to error if these items are not available, not utilized for every fraction, or are misused. The purpose of this project is to design a set of site-specific treatment tolerance tables to be applied to the treatment couch for use in a record and verify (R&V) system that will insure accurate patient setup with minimal workflow interruption. This project also called for the construction of a simple indexing device to help insure reproducible patient setup for patients that could not be indexed with existing equipment. The tolerance tables were created by retrospective analysis on a total of 66 patients and 1,308 treatments, separating them into five categories based on disease site: lung, head and neck (H&N), breast, pelvis, and abdomen. Couch parameter tolerance tables were designed to encompass 95% of treatments, and were generated by calculating the standard deviation of couch vertical, longitudinal, and lateral values using the first day of treatment as a baseline. We also investigated an alternative method for generating the couch tolerances by updating the baseline values when patient position was verified with image guidance. This was done in order to adapt the tolerances to any gradual changes in patient setup that would not correspond with a mistreatment. The tolerance tables and customizable indexing device were then implemented for a trial period in order to determine the feasibility of the system. During this trial period we collected data from 1,054 fractions from 65 patients. We then analyzed the number of treatments that would have been out of tolerance, as well as whether or not the tolerances or setup techniques should be adjusted. When the couch baseline values were updated with every imaging fraction, the average rate of tolerance violations was 10% for the lung, H&N, abdomen, and pelvis treatments. Using the indexing device, tolerances for patients with pelvic disease decreased (e.g., from 5.3 cm to 4.3 cm longitudinally). Unfortunately, the results from breast patients were highly variable due to the complexity of the setup technique, making the couch an inadequate surrogate for measuring setup accuracy. In summary, we have developed a method to turn the treatment couch parameters within the R&V system into a useful alert tool, which can be implemented at other institutions, in order to identify potential errors in patient setup.