Program Information
Automated Daily Cone-Beam CT Quality Control
E Pearson*, N Becker , D Moseley , D Jaffray , T Tadic , The Princess Margaret Cancer Centre - UHN, Toronto, ON
Presentations
TH-AB-601-3 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: 601
Purpose: Monitoring and quality assurance of LINAC mounted imaging systems is an important, but onerous task in current clinical practice. This involves a number of tests performed at varying intervals. However, routinely acquired daily test images contain useful measures of imaging system performance, despite the phantom not being designed expressly for image quality analysis. We have developed an automated image analysis suite to extract a range of metrics that, combined with the daily sampling frequency, may allow for early detection of imaging system performance degradation or component failure, enabling early intervention and avoiding system down time.
Methods: The QUASAR Penta-Guide Phantom (Modus Medical Devices Inc.) is a cubic phantom of a single material with alignment marks and internal spherical voids designed for daily verification of light-field, laser, kV and MV system coincidence. We have developed an automated image processing suite which extracts a range of metrics including measures of geometry, spatial resolution, noise, CT number accuracy and image uniformity. While this design may not be ideal for evaluation of absolute system performance it provides a good relative standard for analysis of trends over time using statistical process control. With this tool, we retrospectively analyzed approximately 500 daily setup data from each of two clinical machines.
Results: For each machine the date range analyzed contained required machine service for kV tube replacement due to observed faults. In each case the SPC analysis showed that extracted metrics from the image analysis responded to the system change and deviated from the normal operating range 1 to 2 months prior to the required service.
Conclusion: A statistical analysis of metrics extracted from images routinely acquired daily can be useful in monitoring system performance and may provide early indication of system failure allowing maintenance to be scheduled to minimize down time and avoid unexpected costs.
Funding Support, Disclosures, and Conflict of Interest: D. Jaffray and D. Moseley have license agreement with Modus Medical Systems
Contact Email: