Author: Emily Hansen, Tyler Kaulfers, Alois M. Ndlovu, Francia Romero, Niara Rowe, Roland Teboh 👨🔬
Affiliation: Montefiore Einstein University Hospital, Hackensack University Medical Center 🌍
Purpose: Daily use and reliance on the auto registration feature requires a performance check. Clinical application of auto registration is performed without a selected region of interest (ROI), requiring a wide range of capabilities. A daily manual quality assurance check of the system’s ability to recenter after a known offset is performed with a 1mm tolerance.
Methods: The 2D diode detector array used for daily quality assurance is manually shifted +2mm, +5mm, +10mm, and -10mm. The radiation therapist registers the images onto the reference image taken during commissioning and compares these results to the ones produced with auto registration. For reproducibility, +10mm was set up and autoregistered three times. The couch was reset to zero and 100 monitor units were delivered for manual and auto registration to obtain a beam center.
Results: Manual registration remained within tolerance while auto registration periodically veered outside of tolerance. For +2mm, +10mm, and -10mm auto registration remained in tolerance while +5mm veered outside of tolerance with x,y,z shifts of -5.8mm, -4.1mm, and -6.3mm. Only -10mm beam center exceeded the tolerance with a negligible x coordinate and 1.05cm y coordinate. While performing +10mm offset three times the resulting auto registration shifts for x were -10.2mm, -10.3mm, and -9.6mm. The shifts for y were -9.5mm, -9.1mm, and -9.2mm. The shifts for z were -10.3mm, -10.4mm, and -10.4mm.
Conclusion: Having a radiation therapist perform the manual and auto registrations provided a clinically relevant analysis of the auto registration function due to its reliance on setup. In most cases, it was obvious the therapist would need to fine-tune the auto registration but it proved mostly successful in staying within tolerance and performing without the selection of an ROI.