Three dimensional volume quantification of aortic valve calcification using multislice computed tomography

Abstract

Objective: To assess a new multislice computed tomography (CT) technique for three dimensional quantification of aortic valve calcification volume (3D AVCV) and to study the relation between stenosis and calcification of the aortic valve.

Methods: 50 patients with echocardiographic calcification of the aortic valve underwent two separate ECG triggered multislice CT for quantification of 3D AVCV. The agreement between the two 3D AVCV scores was assessed and 3D AVCV was compared with echocardiographic markers of severity of aortic stenosis.

Results: Overall the level of agreement between the two 3D AVCV scores was excellent (median interscan variability 7.9% (interquartile range 10.1); correlation coefficient, r = 0.99; repeatability coefficient 237.8 mm3 (limits of agreement -393 to 559 mm3)). However, the magnitude of the 3D AVCV did influence the interscan variability. The 3D AVCV correlated closely with the maximal predicted transvalvar gradient (r2 = 0.77) and aortic valve area (r2 = 0.73).

Conclusions: Multislice CT provides a technique for quantifying 3D AVCV that has good reproducibility. There is a close non-linear relation between echocardiographic parameters of severity of valve stenosis and 3D AVCV scores.

Figure 1

Three dimensional quantification of aortic valve calcification volume (3D AVCV) of aortic valves from two patients viewed in oblique axial orientation. (A) Mild calcification of the left coronary cusp (3D AVCV 74 mm³, aortic valve area 2.4 cm²). (B) Severe calcification (3D AVCV 5044 mm³, aortic valve area 0.4 cm²).

Figure 2

Bland-Altman diagram showing the relation between mean 3D AVCV (in mm³) and the difference between the first 3D AVCV and the second 3D AVCV. The mean difference (dotted line) and limits of agreement, which represent twofold the standard deviation of the mean difference (solid line), are shown.

Figure 3

Relation between 3D AVCV (mm³) and severity of aortic valve stenosis, as determined by (A) echocardiographically derived maximal aortic valve gradient (GD) and (C) continuity equation derived aortic valve area (AVA). The relations are non-linear. 3D AVCV, GD, and valve area datasets were log transformed to correct for non-normality, and the resultant linear correlations with r² values are shown (B and D) (p < 0.00001 for both correlations).