Correcting surface coil intensity inhomogeneity improves quantitative analysis of cardiac magnetic resonance images

Abstract

Quantitative analysis of cardiac magnetic resonance (MR) images is important in bringing objectivity in diagnosis of myocardial abnormalities. Prior to quantitative analysis, it is necessary to correct signal intensity inhomogeneity due to the non-uniform surface coil sensitivity profile. We present a method using non-rigid body image warping and polynomial function fitting to correct this intensity bias on imperfectly registered cardiac MR images. The method was validated on normal human MR images and significantly reduced signal variation from 20.0% to 3.9% in regions of normal myocardium. In MR images of acute myocardial infarction in dogs, signal intensity analysis detected edematous myocardium as 35.9% brighter than normal myocardium on T2- weighted images (p=0.002) while control regions of interest on PD-weighted images were uniform within 2.2% (p=NS). The proposed approach effectively corrected surface coil related signal intensity inhomogeneity in imperfect datasets and allowed confident detection of subtle pathophysiological abnormalities.