| OCR Text |
Show THE UNIVERSITY OF UTAH HEALTH SCIENCES LEAP PROGRAM EVAULATION OF LEFT ATRIAL SHAPE VARIATIONS BETWEEN DIFFERENT ATRIAL FIBRILLATION TYPES USING CORRESPONDENCE-POINT BASED STATISTICAL MODELS Jason Jensen, Diana Phung, Nikolina Savic, (Joshua Cates) Comprehensive Arrhythmia Research and Management (CARMA) Center Scientific Computing and Imaging (SCI) Institute University of Utah Background Atrial fibrillation (AF) is a progressive disease that can be categorized by duration of fibrillation episodes and frequency of occurrence as paroxysmal, persistent or permanent. W e expect to detect significant shape variations of the left atrium (LA) between paroxysmal AF, persistent AF, and non-AF (control) patients by using a shape analysis software. Methods Late Gadolinium Enhancement MRI (LGE-MRI) scans were obtained from 10 paroxysmal and 10 persistent AF patients (age and gender matched) and from 10 control patients. A volumetric segmentation of the LA blood pool was created from the MRI data using Corview software. The resulting segmentations were used as inputs for ShapeWorks software, which generated correspondence-point based statistical models of the LA. Values for principle component analysis modes (dimensions along which shapes show the greatest variation) were compared. Hotelling T-tests were performed between each group to determine the significance of shape difference. Results Hotelling t-tests between control and paroxysmal AF patients showed significance at p<0.05 at the first, third and fourth PCA modes. Tests between control and persistent AF patients showed differences were highly significant at the first and second PCA modes. The differences between paroxysmal and persistent AF patients were also highly significant at the first, second and third PCA modes. Other modes in each group showed potential significance, but parallel analysis disqualified them. The first PCA m o d e in each group is representative of size. As a large portion of the variation in both the control versus persistent and paroxysmal versus persistent groups was due to size, other PCA modes with potential significance dropped below the noise level. Conclusion As expected, the morphology of the LA changes significantly with AF progression as observed size changes are consistent with current literature on LA volume change. Such changes could help aid in AF disease diagnosis and predictions. Figure 1: From left to right, resulting average LA shape with correspondence-point based statistical models In control, paroxysmal AF, and persistent AF. Jason Jensen Diana Phung Nikolina Savic Joshua Cates 167 |
