抄録
Medical image based left ventricle cardiovascular flow was simulated by combinations of optical flow based cardio wall velocity estimation and computational fluid dynamics (CFD). Medical images are obtained a part of diagnosis and they are patient specific. To expand its application, images based computational cardio simulation was suggested in this work. Medical images are based on pixels. Hence, it is facile to use an inertial Eulerian grid arrangement. The advantage of the method is absence of grid generation. Cartesian Coordinate system is directly constructed from the image. Another advantage is known for its stability because the mesh is spatially fixed Since, cardiovascular flow is dynamically triggered by heart muscular movement, identical approach was taken. The external geometry of fluid domain was subjected to movement. To Apply the external deformation, which was calculated by an optical flow algorithm Horn-Schunck Method, to CFD calculation process was initiated. The result illustrated large velocity vectors where significant deformation wan confirmed between steps. Since Horn-Schunck method is based the gradient of pixel luminosity, factious wall velocity vectors was observed even though local geometry was in stationary. Hence, Horn-Schunck method considered to be unsuited for accuracy.
