Quantification and validation of left ventricular wall thickening by a three-dimensional volume element magnetic resonance imaging approach

We have developed a method to quantify and map regional wall thickening throughout the left ventricle (LV) with magnetic resonance imaging. In contrast to methods that measure planar wall thickness and thickening, this method uses the three-dimensional (3D) geometry of the left ventricle to calculate the perpendicular thickness of the wall. We tested this method at three levels of increasing complexity using 1) phantom studies, 2) in vivo experiments in dogs with normal cardiac function, and 3) in vivo studies in dogs during acute ischemia. Experiments were conducted in 15 open-chest dogs imaged by a 0.38 T iron core magnet. Five short-axis images at end diastole and end systole were obtained with the spin echo technique by use of the QRS as a trigger for end diastole and the second heart sound, S₂, to time end systole. After acquisition of preischemic images, acute ischemia was induced by either coronary artery ligation (n=5) or intracoronary dental rubber injection (H=5), which produced severe transmural ischemia. By use of computer-aided contouring of the endocardial and epicardial borders, each image was divided into 16 segments with radial lines originating from the midwall centroid. A 3D volume element was defined as that generated by connecting two matched planar segments in two adjacent image planes. This defined 64 volume elements comprising the entire left ventricle. Thickness and thickening before and during ischemia were then calculated by using the planar segments and the 3D volume elements. In phantom studies, the 3D method was accurate, independent of the angle of inclination of the image plane to the phantom wall, whereas the planar method showed considerable overestimation of thickness when the image plane was oblique to the phantom wall. In the dogs before induction of ischemia, the 3D method demonstrated the well-established normal taper in end-diastolic wall thickness from 1.10±0.02 cm at the base to 1.05±0.11 cm at the apex (p<0.01). By contrast, the planar method did not detect the decrease in thickness toward the apex (1.13±0.07 cm at the base vs. 1.16±0.14 cm at the apex, p=NS). During acute ischemia, thickening was calculated by both methods at the center of the ischemic zone defined by Monastral blue nonstaining and compared with the preischemic values. There was no overlap between the baseline and ischemic values of percent thickening with the 3D method (38.7±14.1% [range, 17-85%] vs. -12.8±13.0% [range, -40% to +13%], p<0.001), whereas there was considerable overlap (34 of 80 regions) with the planar method (35.5±18% [range, 3-75%] vs. -6.3±15.6% [range, -35% to +40%], p<0.001). Finally, a mathematical error model was developed that confirms that the planar thickness method is biased and that this bias is corrected by the 3D method.