JACR: Automated rad dose tool improves quality, builds knowledge
 = |
Although the medical community continues to debate the actual carcinogenic effect of radiation dose, it is clear “that increasing awareness of healthcare professionals regarding imaging-related radiation dose is integral to improving patient care,” wrote Tessa S. Cook, MD, PhD, of the radiology department of the Hospital of the University of Pennsylvania in Philadelphia, and colleagues.
The authors referred to wide variability in estimated dose at the same institution with the same protocols and differences between dose estimates provided in literature and actual reported doses. They cited positive developments including the DICOM Structured Reporting (Dose SR) standard, which stores dose information and provides a platform for the Integrating Healthcare Enterprise program to standardize dose reporting among vendors and the American College of Radiology (ACR) Dose Index Registry Program, a work in progress that will allow direct from the scanner dose reporting.
However, current efforts are not designed to handle retrospective CT data that store dose as an image-based dose sheet, explained the researchers.
Cook and colleagues built an automated extraction pipeline that queries PACS to pull such dose data. The system uses an open-source character recognition tool to translate dose data into ASCII text. Then the DICOM header is translated into an extensible markup language file and an internal script extracts dose parameters. The system also incorporates additional data such as exam data, scanner model, radiologist, technologist and modality section. “The database can be queried by the pathology and radiology enterprise search tool … to enable the real-time generation of patient dose profiles.”
The authors applied the system to all CT studies during the first quarter of this year, and a subset of historical exams since 2003. The automated pipeline allowed the department to analyze multiple factors, including:
- Studies exceeding a dose threshold;
- Estimated dose by scanner and study types, which was used to identify potential differences in protocols that might lead to increased dose;
- Surveys of estimated radiation dose for specific studies, which provide a basis for comparing institutional data with published data; and
- Modality-specific dose estimates.
The department also created radiation dose report cards for technologists, radiologists and referring physicians, which serve as quality assurance measures and educational tools, according to the authors. “Radiologists can use this information to implement dose-reduction protocols, while nonradiologist referring physicians can make informed choices about the type of imaging study to order for a patient’s condition,” wrote Cook. Patient dose report cards provide data about estimated lifetime radiation dose at the hospital.
Cook pointed out that the automated dose extraction mechanism could allow additional analysis and departmental planning. For example, it might enable an evaluation of specific dose reduction measures or inform protocol revision to further decrease dose for certain studies.
The researchers acknowledged that the automated system does not accurately quantify the effective whole-body dose for a specific patient for a given study. That's because CT dose sheets used to approximate patient dose are based on phantom measurements rather than specific patient measurements.
“The automated extraction pipeline for radiation dose information allows us to be complaint with the ACR’s reporting guidelines and to be more cognizant of radiation dose to our patients, thus resulting in improved patient care and management,” concluded Cook and colleagues.