Automatic quantification and characterization of coronary atherosclerosis with computed tomography coronary angiography: cross-correlation with intravascular ultrasound virtual histology
Michiel A de Graaf,Alexander Broersen,Pieter H Kitslaar,Cornelis J Roos,Jouke Dijkstra,Boudewijn P F Lelieveldt,J Wouter Jukema,Martin J Schalij,Victoria Delgado,Jeroen J Bax,Johan H C Reiber,Arthur J Scholte
DOI: https://doi.org/10.1007/s10554-013-0194-x
Abstract:Plaque constitution on computed tomography coronary angiography (CTA) is associated with prognosis. At present only visual assessment of plaque constitution is possible. An accurate automatic, quantitative approach for CTA plaque constitution assessment would improve reproducibility and allows higher accuracy. The present study assessed the feasibility of a fully automatic and quantitative analysis of atherosclerosis on CTA. Clinically derived CTA and intravascular ultrasound virtual histology (IVUS VH) datasets were used to investigate the correlation between quantitatively automatically derived CTA parameters and IVUS VH. A total of 57 patients underwent CTA prior to IVUS VH. First, quantitative CTA quantitative computed tomography (QCT) was performed. Per lesion stenosis parameters and plaque volumes were assessed. Using predefined HU thresholds, CTA plaque volume was differentiated in 4 different plaque types necrotic core (NC), dense calcium (DC), fibrotic (FI) and fibro-fatty tissue (FF). At the identical level of the coronary, the same parameters were derived from IVUS VH. Bland-Altman analyses were performed to assess the agreement between QCT and IVUS VH. Assessment of plaque volume using QCT in 108 lesions showed excellent correlation with IVUS VH (r = 0.928, p < 0.001) (Fig. 1). The correlation of both FF and FI volume on IVUS VH and QCT was good (r = 0.714, p < 0.001 and r = 0.695, p < 0.001 respectively) with corresponding bias and 95 % limits of agreement of 24 mm(3) (-42; 90) and 7.7 mm(3) (-54; 70). Furthermore, NC and DC were well-correlated in both modalities (r = 0.523, p < 0.001) and (r = 0.736, p < 0.001). Automatic, quantitative CTA tissue characterization is feasible using a dedicated software tool. Fig. 1 Schematic illustration of the characterization of coronary plaque on CTA: cross-correlation with IVUS VH. First, the 3-dimensional centerline was generated from the CTA data set using an automatic tree extraction algorithm (Panel I). Using a unique registration a complete pullback series of IVUS images was mapped on the CTA volume using true anatomical markers (Panel II). Fully automatic lumen and vessel wall contour detection was performed for both imaging modalities (Panel III). Finally, fusion-based quantification of atherosclerotic lesions was based on the lumen and vessel wall contours as well as the corresponding reference lines (estimate of normal tapering of the coronary artery), as shown in panel IV. At the level of the minimal lumen area (MLA) (yellow lines), stenosis parameters, could be calculated for both imaging techniques. Additionally, plaque volumes and plaque types were derived for the whole coronary artery lesion, ranging from the proximal to distal lesion marker (blue markers). Fibrotic tissue was labeled in dark green, Fibro-fatty tissue in light green, dense calcium in white and necrotic core was labeled in red.