Meeting Banner
Abstract #3952

Review and Analytical Discussion of Cardiac Segmentation

Esmeralda Ruiz Pujadas1, Marco Reisert1

1Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany, Germany

Segmentation of the myocardium gives valuable information for the the diagnosis and treatment in cardiac pathologies. Manual segmentation is a very tedious and time consuming task. Therefore, an automatic procedure is desirable. But the myocardium detection is very challenging due to the low heterogeneous and discontinuous myocardium, the presence of papillary muscles or the tissues surrounding the myocardium. For that reason, the problem remains open. We will make an overview of the methods applied in medical segmentation and discuss their limitations. After that, an analysis of the methods will be presented.

Keywords

active alignment ambiguity among analytical another appearance approximating around atlas attract authors automated best better boundary built cardiac certain combine commonly complementing compressed constraint constraints consuming contour contours contrast control corners creation curve dealing dependencies dependent described describes detection diagnosis difficulty directional distance edge edges energy error every evolution examples external find flow force forces form function functions gradient graph growing help include incorporated infarction influence initialization intensities intensity inter internal intra issues limitation local long lynch manual measure measures medical merits midas minimize model models motion multistage muscles myocardial myocardium noise normalized novel objects open optimal optimization papillary parametric partition performance placed poor principal prior problem pros registration relies representation representing required review rigid scale search segmentation segmented sensitivity separation shape slice smoothness snakes solve space statistical still strengths structures substantial surface target task thickening thresholding training trans translation update user valuable variability variants ventricular wall watershed whereas whose wide