Comparison of topogram-based body size indices for CT dose consideration and scan protocol optimization

Purpose: To retrospectively compare different topogram-based patient body size indices and to determine the optimal topogram-based body size index as a basis for body computed tomography (CT) dose consideration and scan protocol optimization. Methods: Forty-three routine thorax and abdomen CT scans are studied retrospectively, with patient ages ranging from 18 to 67 yr. The individual patient’s water-equivalent diameter (D w ) of the scanned body region is computed from CT DICOM images as the “gold standard,” after first converting from Hounsfield units values to μa values, where μ is the normalized tissue attenuation coefficient and a is the area per pixel. Four topogram-based body size indices [average diameter ( D ¯ ) , girth (G), topogram projection area (E topo), and improved topogram projection area ( E topo ' ) ] are computed and correlated with D w using linear regression analysis. Specifically, D ¯ is calculated by averaging the coronal and sagittal diameters; G is computed by modeling the patient’s cross-section as an ellipse; E topo is the product of the mean topogram pixel value and the width of the scanned body region; and ( E topo ' ) incorporates E topo with correction of patient miscentering and water attenuation coefficient. The accuracy of these four approaches for estimation of D w is assessed using linear regression models. Results are given in terms of 95% confidence intervals (CIs). Results: Regression analysis results in four different linear models. The standard error (95% CI) for estimation ofD w from D ¯ and G was ±2.8 and ±3.1 cm, respectively (p = 0.297). The standard error for estimation of D w from E topo was significantly less than that from D ¯ (±2.1 cm, p