Subjects
This study was performed retrospectively on CT data acquired as part of a clinical trial protocol investigating the use of CT in inflammatory bowel disease (ClinicalTrials.gov Identifier: NCT 01244386) [5] with approval from the institution Clinical Research Ethics Committee. Fifty adult patients were included in this study and, as part of this trial protocol, all patients signed informed consent.
CT scan protocol
All patients underwent a CT scan of the abdomen and pelvis with a standardised protocol using the following parameters: scan range encompassing the lung bases to the pubic symphysis; 0.625-mm slice acquisition thickness; intravenously administered contrast (Iohexol, Omnipaque 300, General Electric Healthcare, Waukesha, WI, USA) delivered at 2.5 mL/s and imaged in the portal venous phase; 1.5 L of positive oral contrast (2% Gastrografin, Bracco Diagnostics Inc., Princeton, NJ, USA); tube voltage of 120 kVp; automated tube current modulation resulting in a variable current with a minimum of 50 mA and a maximum of 350 mA; gantry rotation time of 0.8 s; noise index 38. All CT images were acquired using a single 64-slice multi-detector row CT scanner (Lightspeed VCT-XTe, GE Healthcare, General Electric Medical Systems, Waukesha, WI, USA). The DLP and CTDIvol values, as well as the corresponding phantom size, were recorded from each CT dose report. CTDIvol and DLP tolerances were verified using a standard 32-cm Perspex phantom, a 10-cm ionisation chamber with a Victoreen NERO mAx unit (Fluke Biomedical, Solon, OH, USA).
The SSDEs were calculated by multiplying the CTDIvol of each patient by conversion factors corresponding to the effective patient diameters in the AAPM reference tables [3]. The imaging performance and assessment from CT patient dosimetry calculator (ImPACT version 0.99x, London, UK) was used to calculate the effective dose.
BMI measurement
Each patient had weight and height measurements performed and their BMI calculated immediately prior to CT scan using a dedicated calibrated measuring device (electronic measuring station Model 763, Seca Medical, Hamburg, Germany). BMI data were used to subdivide patient groups, where underweight referred to BMI < 18.5 kg/m2, normal weight referred to 18.5 ≤ BMI < 25 kg/m2, overweight referred to 25 ≤ BMI < 30 kg/m2 and obese referred to BMI ≥ 30 kg/m2.
Body diameter measurements
Images were reviewed on a picture-archiving and communication system (PACS) workstation (Impax 6.3.1, AGFA Healthcare, Morstel, Belgium) in a DICOM format. As per AAPM Report 204 guidelines, body diameters were measured at the midslice level (median image of the craniocaudal scanning length) on the CT-localiser images because, for larger patients, maximum skin-to-skin distance is often not included on transverse CT images [3, 6]. Diameter measurements were performed manually with the electronic callipers available on the PACS using a window width of 350 Hounsfield Units (HU) and window level of 50 HU. From personal experience at our institution, analysis of interoperator variability for PACS-based anthropometric measurements shows no statistically significant differences. Therefore, a single investigator carried out all measurements. A fixed window level and setting was used for each individual study.
Maximum skin-to-skin anteroposterior diameter (DAP) and lateral diameter (DLAT) were measured in centimetres on lateral and anteroposterior localiser images, respectively. DAP is defined as the anteroposterior skin-to-skin diameter on the lateral localiser at the midslice level (Fig. 1a) while DLAT is defined as the lateral skin-to-skin diameter on the anteroposterior localiser image at the midslice level (Fig. 1b). The DE is defined as the diameter of the circle with area equivalent to the cross-sectional area of the patient at the particular z-axis level (i.e. the midslice level) and is calculated as the geometric mean of DAP and DLAT, as follows:
$$ {D}_E=\surd \left({D}_{AP}\times {D}_{LAT}\right) $$
The outer DE (DOUT) equates to the conventional DE calculated using the AAPM method described above. The inner DE (DIN) is derived using the anteroposterior (DAP(IN)) and lateral diameters (DLAT(IN)) measured on an axial CT image at the midslice level, excluding the subcutaneous adipose tissue (Fig. 1c). The DIN is then calculated as the geometric mean of DAP(IN) and DLAT(IN), as follows:
$$ {D}_{IN}=\surd \left({D}_{AP(IN)}\times {D}_{LAT(IN)}\right) $$
The DE ratio (DRATIO), another patient size-related metrics proposed by Lamoureux et al. [7], was also calculated, as follows:
$$ {D}_{RATIO}={D}_{OUT}/{D}_{IN} $$
Statistical analysis
Data were collated using Microsoft Excel 2010 (Microsoft Corporation, Redmond, WA, USA) and statistical analyses were conducted by using Microsoft Excel 2010 and GraphPad Prism version 5.0 (GraphPad Software Inc., San Diego, CA, USA). Descriptive statistics including means, standard deviations and ranges were calculated. BMI, dose indices (CTDIvol, DLP, SSDE, effective dose) and body diameters (DAP, DLAT, DAP+DLAT, DE, DRATIO) were recorded for each patient CT examination. The correlations between BMI, dose indices, and body diameter measurements were examined with Pearson correlation analysis (r). Linear regression models were used to assess the dependence of CTDI, DLP, SSDE, and effective dose on BMI. Linear regression models were also used to estimate the relationship of effective diameter (independent variable) with BMI (dependent variable). A p-value lower than 0.05 was taken to indicate statistical significance.