Francesco Ria1,2, Wanyi Fu3, Jocelyn Hoye3, W Paul Segars3, Anuj J Kapadia3, Ehsan Samei3,4,5. 1. Carl E. Ravin Advanced Imaging Labs, Duke University Health System, 2424 Erwin Road, Suite 302, Durham, NC, 27710, USA. francesco.ria@duke.edu. 2. Clinical Imaging Physics Group, Duke University Health System, 2424 Erwin Road, Suite 302, Durham, NC, 27710, USA. francesco.ria@duke.edu. 3. Carl E. Ravin Advanced Imaging Labs, Duke University Health System, 2424 Erwin Road, Suite 302, Durham, NC, 27710, USA. 4. Clinical Imaging Physics Group, Duke University Health System, 2424 Erwin Road, Suite 302, Durham, NC, 27710, USA. 5. Medical Physics Graduate Program, Departments of Radiology, Physics, Biomedical Engineering, and Electrical and Computer Engineering, Duke University, 2424 Erwin Road, Suite 302, Durham, NC, 27710, USA.
Abstract
OBJECTIVES: Quantifying radiation burden is essential for justification, optimization, and personalization of CT procedures and can be characterized by a variety of risk surrogates inducing different radiological risk reflections. This study compared how twelve such metrics can characterize risk across patient populations. METHODS: This study included 1394 CT examinations (abdominopelvic and chest). Organ doses were calculated using Monte Carlo methods. The following risk surrogates were considered: volume computed tomography dose index (CTDIvol), dose-length product (DLP), size-specific dose estimate (SSDE), DLP-based effective dose (EDk ), dose to a defining organ (ODD), effective dose and risk index based on organ doses (EDOD, RI), and risk index for a 20-year-old patient (RIrp). The last three metrics were also calculated for a reference ICRP-110 model (ODD,0, ED0, and RI0). Lastly, motivated by the ICRP, an adjusted-effective dose was calculated as [Formula: see text]. A linear regression was applied to assess each metric's dependency on RI. The results were characterized in terms of risk sensitivity index (RSI) and risk differentiability index (RDI). RESULTS: The analysis reported significant differences between the metrics with EDr showing the best concordance with RI in terms of RSI and RDI. Across all metrics and protocols, RSI ranged between 0.37 (SSDE) and 1.29 (RI0); RDI ranged between 0.39 (EDk) and 0.01 (EDr) cancers × 103patients × 100 mGy. CONCLUSION: Different risk surrogates lead to different population risk characterizations. EDr exhibited a close characterization of population risk, also showing the best differentiability. Care should be exercised in drawing risk predictions from unrepresentative risk metrics applied to a population. KEY POINTS: • Radiation risk characterization in CT populations is strongly affected by the surrogate used to describe it. • Different risk surrogates can lead to different characterization of population risk. • Healthcare professionals should exercise care in ascribing an implicit risk to factors that do not closely reflect risk.
OBJECTIVES: Quantifying radiation burden is essential for justification, optimization, and personalization of CT procedures and can be characterized by a variety of risk surrogates inducing different radiological risk reflections. This study compared how twelve such metrics can characterize risk across patient populations. METHODS: This study included 1394 CT examinations (abdominopelvic and chest). Organ doses were calculated using Monte Carlo methods. The following risk surrogates were considered: volume computed tomography dose index (CTDIvol), dose-length product (DLP), size-specific dose estimate (SSDE), DLP-based effective dose (EDk ), dose to a defining organ (ODD), effective dose and risk index based on organ doses (EDOD, RI), and risk index for a 20-year-old patient (RIrp). The last three metrics were also calculated for a reference ICRP-110 model (ODD,0, ED0, and RI0). Lastly, motivated by the ICRP, an adjusted-effective dose was calculated as [Formula: see text]. A linear regression was applied to assess each metric's dependency on RI. The results were characterized in terms of risk sensitivity index (RSI) and risk differentiability index (RDI). RESULTS: The analysis reported significant differences between the metrics with EDr showing the best concordance with RI in terms of RSI and RDI. Across all metrics and protocols, RSI ranged between 0.37 (SSDE) and 1.29 (RI0); RDI ranged between 0.39 (EDk) and 0.01 (EDr) cancers × 103patients × 100 mGy. CONCLUSION: Different risk surrogates lead to different population risk characterizations. EDr exhibited a close characterization of population risk, also showing the best differentiability. Care should be exercised in drawing risk predictions from unrepresentative risk metrics applied to a population. KEY POINTS: • Radiation risk characterization in CT populations is strongly affected by the surrogate used to describe it. • Different risk surrogates can lead to different characterization of population risk. • Healthcare professionals should exercise care in ascribing an implicit risk to factors that do not closely reflect risk.
Authors: Francesco Ria; Joseph T Davis; Justin B Solomon; Joshua M Wilson; Taylor B Smith; Donald P Frush; Ehsan Samei Journal: AJR Am J Roentgenol Date: 2019-06-10 Impact factor: 3.959
Authors: Kalpana M Kanal; Priscilla F Butler; Debapriya Sengupta; Mythreyi Bhargavan-Chatfield; Laura P Coombs; Richard L Morin Journal: Radiology Date: 2017-02-21 Impact factor: 11.105
Authors: E Vañó; D L Miller; C J Martin; M M Rehani; K Kang; M Rosenstein; P Ortiz-López; S Mattsson; R Padovani; A Rogers Journal: Ann ICRP Date: 2017-10
Authors: Ehsan Samei; Hannu Järvinen; Mika Kortesniemi; George Simantirakis; Charles Goh; Anthony Wallace; Eliseo Vano; Adrian Bejan; Madan Rehani; Jenia Vassileva Journal: J Radiol Prot Date: 2018-05-17 Impact factor: 1.394