Automated Detection of IGRT Shift Errors Due to Mis-Registration
D Pittock1*, A Kruzer1, D Faught1, A Nelson1, E Paulson2, (1) MIM Software Inc., Cleveland, OH, (2) Medical College of Wisconsin, Milwaukee, WI
Presentations
PO-GePV-M-147 (Sunday, 7/25/2021) [Eastern Time (GMT-4)]
Purpose: Critical errors can result from incorrect shifts being applied during Image-Guided Radiation Therapy (IGRT). Incorrect shifts can result from mis-registering cone-beam CT (CBCT) images to reference images, or failing to move the linac couch by the intended shift. Once couch coordinates are captured, tolerance tables can be effective at identifying IGRT shift errors due to mis-registration. However, if the initial couch coordinates are captured incorrectly during verification simulation, a systematic error in the treatment could result. We demonstrate here an automated method to detect IGRT shift errors due to mis-registrations.
Methods: Daily CBCT images and spatial registration object files containing actual IGRT shifts for six patients were transferred to MIM. Residual shifts were determined by the difference between actual IGRT shifts and an automated rigid registration performed by MIM ART Assist. To determine an action level for residual shift errors, treatment plans were recalculated on daily CBCT images with isocenter shifts of +/- 0.5, 1.0, 2.0, and 5.0 cm applied along the three cardinal directions using MIM SureCalc ART. Automated detection of residual shift errors was then implemented clinically in MIM ART Assist.
Results: For shift errors of +/- 0.5 cm, no difference in target coverage was detected. For shift errors of +/- 1.0 cm, target coverage fell to 94% on average (range 92-97%). Beyond +/- 1.0 cm, target coverage fell below 80%. Based on these results, a residual shift error action level of +/- 1.0 cm was employed. During clinical use over 323 fractions, 8 fractions (3 chest, 1 abdomen, 4 pelvis patients) demonstrated residual shift errors exceeding +/- 1.0 cm that required further investigation.
Conclusion: We have demonstrated an automated method to detect IGRT shift errors due to mis-registration. A residual shift error action level of +/- 1.0 cm may be appropriate to trigger further investigation.
ePosters
Keywords
Not Applicable / None Entered.
Taxonomy
IM/TH- Image Registration: General (Most aspects)
Contact Email