Correction of lumen contrast-enhancement influence on non-calcified coronary atherosclerotic plaque quantification on CT.

Abstract

Lumen contrast-enhancement influences non-calcified atherosclerotic plaque Hounsfield-unit (HU) values in computed tomography (CT). This study aimed to construct and validate an algorithm to correct for this influence. Three coronary vessel phantoms with 1, 2, and 4 mm circular hollow lumina; with normal and plaque-infested walls were scanned simultaneously in oil using a dual-source CT scanner. Scanning was repeated as the lumina were alternately filled with water and four contrast solutions (100-400 HU, at 100 HU intervals). Images were reconstructed at 0.4 mm x-y pixel size. Pixel-by-pixel comparisons of contrast-enhanced and non-contrast-enhanced images confirmed exponential declining patterns in lumen contrast-enhancement influence on wall HU-values from the lumen border (y = Ae(-λx) + c). The median difference of the inside and outside 2-pixel radius part of the contrast-enhanced coronary phantom wall to the reference (non-contrast-enhanced images) was 45 and 2 HU, respectively. Based on the lumen contrast-enhancement influence patterns, a generalized correction algorithm was formulated. Application of the generalized correction algorithm to the inside 2-pixel radius part of the wall reduced the median difference to the reference to 4 HU. In conclusion, lumen contrast-enhancement influence on the vessel wall can be defined by an exponential approximation, allowing correction of the CT density of the vessel wall closest to the lumen. With this correction, a more accurate determination of vessel wall composition can be made.