Precision and accuracy in CT attenuation measurement of vascular wall using region-of-interest supported by differentiation curve

Objectives

To evaluate the precision and accuracy in CT attenuation measurement of vascular wall using region-of-interest (ROI) supported by differentiation curves.

Study design

We used vascular models (actual attenuation value of the wall: 87 HU) with wall thicknesses of 1.5, 1.0, or 0.5 mm, filled with contrast material of 250, 348, or 436 HU. The nine vascular models were scanned with a 64-detector CT. The wall attenuation values were measured using three sizes (diameter: 0.5, 1.0, and 1.5 mm) of ROIs without differentiation curves. Sixteen measurements were repeated for each vascular model by each of two operators. Measurements supported by differentiation curves were also performed. We used analyses of variance with repeated measures for the measured attenuations for each size of the ROI.

Results

Without differentiation curves, there were significant differences in the attenuation values of the wall among the three densities of contrast material, and the attenuation values tended to be overestimated more as the contrast material density increased. Operator dependencies were also found in measurements for 0.5- and 1.5-mm thickness models. With differentiation curves, measurements were not possible for 0.5- and 1.0-mm thickness models. Using differentiation curves for 1.5-mm thickness models with a ROI of 1.0- or 1.5-mm diameter, the wall attenuations were not affected by the contrast material densities and were operator independent, measuring between 75 and 103 HU.

Conclusions

The use of differentiation curves can improve the precision and accuracy in wall attenuation measurement using a ROI technique, while measurements for walls of ≤1.0 mm thickness are difficult.