Using High Resolution Cardiac CT Data to Model and Visualize Patient-Specific Interactions Between Trabeculae and Blood FlowScott Kulp1, Mingchen Gao1, Shaoting Zhang1, Zhen Qian2, Szilard Voros2, Dimitris Metaxas1 and Leon Axel3

1 CBIM Center, Rutgers University, 2Piedmont Heart Institute, 3New York University

This material is based upon work supported by the U.S. Department of Homeland Security under Grant Award Number 2007-ST-104-000006. The views and conclusions contained in this document are those of the authors and should not be interpreted as necessarily representing the official policies, either expressed or implied, of the U.S. Department of Homeland Security.

Contact email:sckulp@cs.rutgers.edu

Introduction• After a heart attack, the movement of the heart walls changes, affecting the motion of blood. This could potentially lead to thrombus and stoke.• While imaging techniques such as ultrasound and MRI can monitor some blood flow, the image resolutions are low and cannot capture the interactions between the highly complex heart walls and the blood.• We, instead, seek methods to model the patient-specific structure and motion of the detailed geometry of the heart walls to the simulate the flow of blood in the left ventricle to assist in diagnosis.

Data Acquisition

Model Generation

• To model the geometry of the heart, we use CT images from a 320-MSCT scanner (Toshiba Aquilion ONE) using contrast agent.• This advanced diagnostic imaging system captures the whole-heart scan in a single rotation, and achieves 0.3mm volumetric resolution. • 3D+time CT data was acquired in ten frames in a single heart beat cycle, and had an in-plane dimension of 512x512x320 pixels.

• To generate the 3D mesh from data, we use snake based semi-automatic segmentation to acquire the initial segmentation for the first frame of data.• The initial mesh is generated as an isosurface of the segmentation, which we deform to match the shape of the heart at each consecutive frame, in order to achieve the necessary one-to-one correspondence of vertices between frames.• Reconstruction results for a healthy and diseased heart achieve high levels of structural detail that have never been simulated before.

Visualizing Average Residency Time• At the initial time step, particles are generated randomly within the heart. At the beginning of each consecutive time step, new particles are generated at the mitral valve, allowing fresh blood particles to enter the heart during diastole.• We then use simple Eulerian time integration to move each particle according to the fluid velocity at every time step. We can use this to measure and visualize the average age of blood particles (blue=new, green=medium, red=old), revealing how blood may become trapped within the trabeculae in abnormal hearts.

Slowed heart: Blood is trapped within parts of the trabeculae

Dyssynchronous heart: Blood is very poorly circulated, not

moving out of trabeculae

Normal heart: Blood is well-circulated, very few red regions.

Fluid Simulation

Streamlines show blood entering trabeculae during

diastole

Streamlines show blood entering trabeculae during

diastole

• To simulate blood flow through the heart, we represent the 3D meshes as Marker Level Sets and use Finite Difference Method to solve the Navier-Stokes equations:

0

2

u

uPuutu

Healthy Heart

Diseased Heart