The Medical Physics 3.0 initiative aims to enhance direct physicist involvement in
clinical decision making to improve patient care. In this involvement, it is crucial
to achieve effective and patient-specific radiation risk assessment. CT risk characterization
presents a variety of metrics, many of which used as radiation risk surrogates; some
are related to the device output (CTDI), whereas others include patient organ risk-,
age-, and gender-factors (Effective Dose, Risk Index). It is unclear how different
metrics can accurately reflect the radiological risk. This study compared how twelve
metrics characterize risk across CT patient populations to inform effective clinical
decision making in radiology.
This IRB-approved study included 1394 adult CT examinations (abdominopelvic and chest).
Organ doses were calculated using Monte Carlo methods. The following risk surrogate
metrics were calculated: CTDIvol, DLP, SSDE, DLP-based Effective Dose (EDk), organ-dose-based
ED (EDOD), dose to defining organ (stomach- and lungs-ODD), organ-dose-based Risk
Index (RI), and 20 y.o. patient Risk Index (RIr). Furthermore, ODD,0, ED0, and RI0
were calculated for a reference patient (ICRP 110). Lastly, an adjusted ED (ED') was
computed as the product of RI/RIr and EDOD. A linear regression was applied to assess
each metric’s dependency to RI, assumed to be the closest patient risk surrogate.
The normalized-slope (nS) and a Minimum Risk Detectability Index (MRDI=RMSE/slope)
were calculated for each fit.
The analysis reported significant differences between the metrics. ED’ showed the
best concordance with RI in terms of nS and MRDI. Across all metrics and protocols,
nS ranged between 0.37(SSDE) to 1.29(RI0); MRDI ranged between 39.11(EDk) to 1.10(ED’)
cancers per 105 patients per 0.1Gy.
Radiation risk characterization in CT populations is strongly affected by the index
used to describe it. When involved in clinical decisions, medical physicists should
exercise care in ascribing an implicit risk to factors that do not closely reflect
Published Version (Please cite this version)10.1002/mp.14316