Total adipose area measurements (VAA + SAA) obtained from a single slice (L4 L5 + 6 cm) correlated very closely with total adipose volume (Spearman's correlation r = 0.978, p < 0.001).
Stepwise linear regression using DLP as the dependent and BMI and total adipose tissue as independent variables demonstrated that total adipose tissue is more predictive of DLP than BMI [B (95% CI) = 16.045 (11.337-20.752), t=6.681, p < 0.001].
Agreement was best for total adipose volume estimation between observers (intraclass correlation coefficient of TAT = 0.999, p < 0.001, n = 71).
However, obtaining a total adipose volume would also be impossible in patients who are undergoing their first CT scan.
Visceral adipose area (VAA) and subcutaneous adipose area (SAA) were measured on a single slice so that these could be compared with total adipose volume estimates.
Total adipose volume was again found to be the strongest predictor of dose length product (DLP) in each individual age-sex subgroup.
Patient dose length product was the dependent variable and total adipose, muscle, bone, liver, kidney and spleen volumes were included as independent variables in the regression model.
We found that total adipose volume was the strongest predictor and that muscle volume was also a significant but weaker predictor of dose length product.
We found that BMI correlated positively and significantly with all measured body composition parameters, in particular total adipose tissue (Spearman correlation = 0.798, p < 0.001) (Table 4).
A practical limitation of our study is that quantitatively obtaining a patient's total adipose volume represents an additional workflow challenge in the day-to-day practice of CT.
Stepwise linear regression using dose length product as the dependent and BMI and total adipose tissue as independent variables demonstrated however that total adipose tissue [B 95% CI) = 0.022 (0.019 0.025), t = 9.788, p < 0.001] is more predictive of dose length product than BMI [B 95% CI) = 16.045 (11.337–20.752), t = 6.681, p < 0.001].
