Click here to


Are you sure ?

Yes, do it No, cancel

Using In-Vivo EPID Measurements to Detect Deviations and Track Trends in Machine Performance Per Treatment Day

C Bojechko*, J Zavala Bojorquez, University of California San Diego, La Jolla, CA


(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: Daily QA performed prior to treatment is a ubiquitous QA strategy, however it will not detect errors that can occur during the daily operation of the linear accelerator’s (linac). In-vivo measurements provide an extra layer of safety and will detect deviations that occur during clinical operation. We use a novel approach analyzing in-vivo EPID images to detect deviations and track trends in machine performance per treatment day. This method can detect large errors that occur during a linac’s operation as well as any systematic deviations on a given treatment day.

Methods: In-vivo EPID images were collected on a Varian Halcyon, over a 23-month period, for every patient and every fraction of treatment. For each fraction of treatment a mean image difference ( µ_image) was computed, using pixels in the high dose region and the first fraction image as a baseline. A daily average (µ_daily) was computed, averaging the µ_image over all patients treated on the same day.

Results: To reduce noise, only treatment days with greater than 20 patients treated were considered. This resulted in 437 treatment days, the average and standard deviation of µ_daily was 0.04%±0.38%. The machine maintenance records and the daily machine performance check (MPC) were reviewed for the same period of treatment. Abrupt changes (>1.0% in µ_daily) corresponded to machine output adjustments and imager recalibrations. The magnitude of changes in µ_daily agreed with changes from TG-51 output adjustments.

Conclusion: Using in-vivo EPID images averaged over all patients on a treatment day can provide additional machine quality assurance . This retrospective study showed that changes in the machine output and imager recalibration on the order of 1-2% can be detected by tracking a daily treatment average. We now use this technology prospectively, automatically downloading images and calculating µ_daily at the end of the treatment day.

Funding Support, Disclosures, and Conflict of Interest: Funding support from Varian Medical Systems


Not Applicable / None Entered.


TH- External Beam- Photons: portal dosimetry, in-vivo dosimetry and dose reconstruction

Contact Email