A novel method decomposing left ventricular deformation into separate modes successfully distinguished diabetes patients from normal subjects, with an average Euclidean distance error of <1.7 mm.
Observational (n=26)
Can a novel method decomposing left ventricular deformation into separate modes accurately quantify deformation and distinguish between normal subjects and diabetes patients?
A novel computational method using MR tagged images successfully quantifies left ventricular deformation modes and can distinguish abnormal deformation patterns in diabetes patients from normal subjects.
We have developed a method that decomposes the deformation of the left ventricle (LV) between end diastole (ED) and end systole (ES) into separate deformation modes such as longitudinal shortening, wall thickening, and twisting. The deformation was initially found from the motion of an LV finite-element mesh that was fitted to clinically obtained magnetic resonance (MR) tagged images. A mode coefficient was calculated for each deformation mode to quantify the different modes and, thus allowing for discrimination of normal and abnormal deformation patterns. We applied the method to 13 normal subjects and 13 diabetes patients. By using the ED mesh as reference and adding the extracted deformation modes multiplied by their mode coefficients, an approximate ES mesh was calculated and compared with the "true" ES mesh found from the MR images. For the 26 subjects the average Euclidean distance was less than 1.7+/-0.9 mm between the nodes of the approximated and true ES meshes. The coefficient values for the patient group showed significantly less longitudinal shortening, less wall thickening, more longitudinal twisting and also more bulging of the septum into the LV when compared with the normal subjects. We conclude that the developed method successfully quantifies the deformation into several modes of deformation and is capable of distinguishing the deformation of a group of patients from a group of normal subjects.
Remme et al. (Tue,) conducted a observational in Diabetes (n=26). Left ventricular deformation mode decomposition method vs. Normal subjects was evaluated on Average Euclidean distance between the nodes of the approximated and true end-systole meshes. A novel method decomposing left ventricular deformation into separate modes successfully distinguished diabetes patients from normal subjects, with an average Euclidean distance error of <1.7 mm.