Graham Ingersoll Warden MD, Presenter: Nothing to Disclose

Aaron D. Sodickson MD, PhD, Abstract Co-Author: Consultant, Siemens AG Consultant, Bayer AG

Cameron Farkas, Abstract Co-Author: Nothing to Disclose

Luciano Monte Serrat Prevedello MD, Abstract Co-Author: Nothing to Disclose

Katherine P. Andriole PhD, Abstract Co-Author: Nothing to Disclose

Ramin Khorasani MD, Abstract Co-Author: Stockholder, Medicalis Corp Advisory Board, General Electric Company Advisory Board, EMC Corp

In order to address established challenges in radiation exposure monitoring, it is necessary to extract anatomy-specific CT exposure metrics from existing image repositories. The purpose of this study is to validate the performance of an open-source tool that automatically extracts this information, and to demonstrate how the resulting data can enhance radiation monitoring efforts.

The IRB approved this retrospective, HIPAA compliant study. We internally developed software (derived from the open source PixelMed DICOM toolkit) that extracts anatomy-specific CT exposure metrics through optical character recognition of dose report screen captures in combination with DICOM attributes. A random sample of 54,548 CT encounters from 2000-2010 was extracted from the enterprise image repository that includes our adult academic tertiary referral hospital, affiliated cancer center, community hospital, outpatient imaging sites, and outside examinations imported from portable media. For toolkit validation, we randomly selected 150 encounters with dose screens for each major CT manufacturer and measured the recall of dose report screen captures and the extraction precision of anatomy-specific exposure metrics.

Dose-event recall by CT encounter was 98.4% (95% CI’s 97.4-99.4%), with numerical exposure metric extraction precision of 100%. Precision of anatomic assignment was 92%, expressed as the mean fraction of the CT encounter’s total dose length product (DLP) with correctly assigned anatomy. Anatomically incorrect or nonspecific protocol and series descriptions accounted for 81% (87/108) of the errors in anatomic assignment. We demonstrate use of the extracted data for i) radiation safety benchmarking within and between institutions, ii) patient- and anatomy-specific radiation dose monitoring, and iii) improved CT protocol quality control.

New open source software tools can successfully perform large-scale automated extraction of anatomy-specific CT exposure data from existing image repositories, thus greatly enhancing radiation monitoring capabilities.

This new open-source tool enables radiation exposure benchmarking and diagnostic reference level development, improved CT quality control, and refined patient-centric longitudinal dose monitoring

Warden, G, Sodickson, A, Farkas, C, Prevedello, L, Andriole, K, Khorasani, R, Exposing Exposure: Automated Anatomy-specific CT Radiation Exposure Extraction from Existing Image Repositories.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11006106.html