Implementation of Size-Dependent Local Diagnostic Reference Levels for CT Angiography

OBJECTIVE. Diagnostic reference levels (DRLs) are established for standard-sized patients; however, patient dose in CT depends on patient size. The purpose of this study was to introduce a method for setting size-dependent local diagnostic reference levels (LDRLs) and to evaluate these LDRLs in comparison with size-independent LDRLs and with respect to image quality. MATERIALS AND METHODS. One hundred eighty-four aortic CT angiography (CTA) examinations performed on either a second-generation or third-generation dual-source CT scanner were included; we refer to the second-generation dual-source CT scanner as "CT1" and the third-generation dual-source CT scanner as "CT2." The volume CT dose index (CTDI(vo)l) and patient diameter (i.e., the water-equivalent diameter) were retrieved by dose-monitoring software. Size-dependent DRLs based on a linear regression of the CTDIvol versus patient size were set by scanner type. Size-independent DRLs were set by the 5th and 95th percentiles of the CTDIvol values. Objective image quality was assessed using the signalto- noise ratio (SNR), and subjective image quality was assessed using a 4-point Likert scale. RESULTS. The CTDIvol depended on patient size and scanner type (R-2 = 0.72 and 0.78, respectively; slope = 0.05 and 0.02 mGy/mm; p < 0.001). Of the outliers identified by size-independent DRLs, 30% (CT1) and 67% (CT2) were adequately dosed when considering patient size. Alternatively, 30% (CT1) and 70% (CT2) of the outliers found with size-dependent DRLs were not identified using size-independent DRLs. A negative correlation was found between SNR and CTDIvol (R-2 = 0.36 for CT1 and 0.45 for CT2). However, all outliers had a subjective image quality score of sufficient or better. CONCLUSION. We introduce a method for setting size-dependent LDRLs in CTA. Size-dependent LDRLs are relevant for assessing the appropriateness of the radiation dose for an individual patient on a specific CT scanner.