Avaliação quantitativa da fibrose miocárdica pela ressonância magnética cardíaca nas cardiomiopatias não isquêmicas utilizando o método semi-quantitativo visual: Comparação com a quantificação por planimetria

Abstract

The assessment of myocardial fibrosis (MF) by CMR using late gadolinium enhancement (LGE) has been shown to provide valuable prognostic information in patients with non-ischemic heart diseases (NIHD). Currently, quantitative assessment of MF requires time-consuming analyses using commercially available dedicated software. We describe a new method of MF quantification using a semi-quantitative visual analysis of LGE images and evaluate its agreement with frequently used thresholding techniques. Contrast-enhanced CMR was performed in 90 patients with NIHD (hypertrophic cardiomyopathy [n=30]; non-ischemic dilated cardiomyopathy [n=30]; and acute myocarditis [n=30]). Only patients with identifiable regions of LGE were included. In order to quantify MF by the semi-quantitative visual analysis technique (Visual), LGE slices were segmented into 44 segments (2 apical slices with 4 segments each, 3 mid-ventricular slices with 6 segments each and 3 basal slices with 6 segments each). The amount of LGE in each segment was scored using a 6-point scale: 0-no LGE; 1->0% and &#8804;5%; 2->5% and &#8804;25%; 3->25% and &#8804;50%; 4->50% and &#8804;75%; and 5->75%. The total amount of LGE was calculated as the sum of the scores of all 8 slices divided by 220 (maximum possible score: 5*44=220) and was reported as a percent of LV mass. MF quantification was also performed using 4 different semiautomated thresholding techniques: 2SD of remote (2SD); 6SD of remote (6SD); manual threshold (Manual); and full width at half maximum (FWHM). The amount of MF quantified by Visual (15.8±9.7%) was similar to that quantified by Manual (15.8±10.5%, p=0.85) and FWHM (16.2±10.1%, p=0.27), was slightly higher than 6SD (14.4±10.5%, p=0.002) and was significantly lower than 2SD (39.6±14.5%, p<0.001). There was good agreement between Visual and all thresholding techniques, except 2SD, which significantly overestimated the amount of LGE. When we evaluated the agreement within each disease subgroup the pattern was similar to that observed for the entire population. Inter- and intraobserver variabilities were small for all measurements. Importantly, the average time/patient for the Visual technique (1 and 35 ) was significantly shorter than Manual (9 and 12 , p<0.001), FWHM (9 and 19 , p<0.001) and 2SD/6SD (9 and 55 , p<0.001). Quantification of MF by the novel semi-quantitative visual technique demonstrated good agreement with standard thresholding methods. Our results indicate that it could represent a faster and more practical alternative for the post-processing analyses of LGE images in patients with NIHD.