Analysis of right ventricular remodeling using curvature histogram comparison

Abstract

Understanding and quantifying right ventricular (RV) remodeling in repaired Tetralogy of Fallot (TOF) is crucial for patient management and therapy planning, i.e., in determining the optimal time for pulmonary valve replacement. However, quantification of RV remodeling is usually hampered by its complex geometry. This paper presents a computer-aided cardiac assessment methodology to quantitate RV remodeling in terms of a histogram similarity index, based on the surface curvature distribution of three-dimensional (3D) RV geometries at both the end-diastole and end-systole phases. These 3D RV geometries are reconstructed from border delineated cardiac MRI images, whereby a surface fitting algorithm is then used to calculate the curvature distribution of the 3D models. The curvature histograms at ED and ES are computed and their similarities are measured using the Bhattacharya Similarity Metric, which is denoted as hdist. Based on an initial study involving 5 TOF patients and 5 normal subjects, we observed that the mean hdist for the normal controls is significantly higher (p = 0.0015 < 0.05 and p′ = 0.004 < 0.05; student Mest and Mann-Whitney-Wilcoxon test, respectively) as compared to that of the TOF patients. This suggests that hdist can be used as a discriminant between TOF patients and normal control.