Room: AAPM ePoster Library
Purpose: To assess prostate positioning accuracy and movement based on fiducial marker positions identified on CBCT and planar kV images.
Methods: To guide accurate patient setup, CBCT images were collected immediately prior to treatment. If triggered on-board kV images showed any of the three fiducials strayed farther than 3 mm from planning-expected positions, the treatment was interrupted, another CBCT of the patient was obtained, and corrective shifts were applied prior to continuing treatment delivery. Retrospectively, the fiducial marker positions were automatically identified on the planning CTs and 101 CBCTs of 15 patients undergoing prostate SBRT. The pelvic bony anatomy in the CBCTs was rigidly registered to the planning CT, and internal motion was calculated based on fiducial positions identified in these registered CBCTs. By comparing internal motion to corrective couch movements (global motion), the extent of external patient motion was identified. The relative prostate size was measured by calculating the area of the triangle formed by the three fiducial markers.
Results: 96.2% of fiducials identified on CBCT were within 3 mm of their expected, planning positions. Of the 50 fractions investigated, treatment was interrupted in 16 cases. All resulting couch corrective shifts were within 2 mm of the identified internal prostate motion, suggesting that the cause of treatment interruption is internal prostate motion, not external patient motion. Between CT and final treatment, the prostate was observed to move internally up to 20 mm. The prostate size varies from 86 to 109% of its original, planning CT size.
Conclusion: Software was developed to automatically identify fiducial positions in CBCT images and register the CBCT to the planning CT based on bony pelvic anatomy. The coincidence of couch motion to internal prostate motion suggests that patients are properly immobilized, and that treatment interruptions are due to internal changes.