Stacy Kellan Goergen MBBS, Presenter: Nothing to Disclose

Anthony Wallace BS, PhD, Abstract Co-Author: Nothing to Disclose

Daniel Schick BSc, PhD, Abstract Co-Author: Nothing to Disclose

Tina Soblusky MS, Abstract Co-Author: Nothing to Disclose

To measure variation in dose delivered by CT scanning and to provide training in optimisation.

10 CT facilities with one CT scanner each (Philips Brilliance 64, Philips Brilliance 16, GE LightSpeed VCT, Toshiba Aquilion 64, Toshiba Aquilion 16, Siemens Sensation 64, Siemens Definition AS+) were surveyed for the scanning parameters used for 4 adult and 8 pediatric CT protocols. Collected scan data included kVp, mAs, collimation, scan length, pitch, use of dose modulation (long axis, rotational, and patient size) age, gender, and girth. Scanner generated CTDIvol and DLP were recorded. Scan parameters were entered into a web based tool and CT Expo software used for dose calculation . Actual patient scan data was collected for all adult protocols. Patient data supplemented with five-year-old phantom data was used for paediatric protocols, due to insufficient actual paediatric patient scans being performed during the audit. Dose length product (DLP) was the indicative metric for dose comparison, supplemented by effective dose for the pediatric protocols. De-identified data was presented to participating sites at a feedback workshop for radiologists and technologists. Dose optimisation strategies were discussed.    

5 to 10 - fold and 4 to 30 - fold variation in dose for adult and pediatric protocols, respectively, was found, unaccounted for by girth variation. Differing effective mAs and kVP and use of dose modulation accounted for most of the variation. Siemens and Philips scanners used the 32cm CTDI phantom model to generate DLP for all non-head scans, in accordance with IEC recommendations, even when these were pediatric small field - of - view scans, resulting in scanner generated DLP being approximately half the CT Expo calculated value. Toshiba and GE scanners varied the phantom model used for DLP calculation dependent upon scan field - of - view, predominantly resulting in higher scanner generated DLP for pediatric non-head protocols.

Variation in CT dose is great and largely accounted for by effective mAs and kVp. Between scanner comparison of manufacturer - generated DLP for non-head examinations may be invalid due to confounding by differences in the method of DLP calculation.

5 - 9 fold variation in radiation dose delivered by CT is unlikely to be clinically warranted. Inter-scanner comparison of dose for non-head pediatric studies may be misleading.

Goergen, S, Wallace, A, Schick, D, Soblusky, T, Measuring CT Dose: Lessons from Optimization.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8003046.html