12.4 Mb avi
Virtual slipstream animation of pulsatile flow in a giant aneurysm CFD model reconstructed from 3D rotational angiography. This is the animation from which we extracted still frames for our paper in Am J Neuroradiol 2003
3.8 Mb avi
Particle path animation of pulsatile flow in a mildly diseased carotid bifurcation CFD model reconstructed from black blood MRI. This is the animation from which we extracted still frames for our paper in Magn Reson Med 2002.
300 kb mpeg
Longitudinal and cross sections of peak systolic velocity at the carotid bifurcation of a post-endarterectomy patient. Although B-mode showed no evidence of stenosis, Doppler spectra showed elevated velocities suggesting stenosis. By simulating the appearance of the color Doppler images using the CFD computed velocity field reconstructed from MRI we aim to understand the source of this conflicting clinical data.
8.1Mb avi
Visualization of pulsatile flow in the vessel shown segmented below. Note that we have truncated the external branch below the level of its bifurcation, for reasons described below. See below for a description of the animation.
2.3Mb mpeg
Carotid bifurcation reconstructed from serial MRI images of a 78 year old volunteer using our 3-D segmentation tools. Such vessels would be difficult, if not impossible, to reconstruct using conventional serial techniques. Note the presence of a small side branch in the external carotid into which the mesh has only just penetrated. Future work is aimed at improving the resolution of the MRI acquisitions to better resolve such side branches for inclusion in the CFD models.
300kb mpeg
These two animations show the results of pulsatile CFD studies in simplified models of the distal end of a coronary bypass graft. In the top animation, the presence of a graft downstream (i.e. to the right) of the stenosis (far left) and branch produces relatively slow flow in the host between the graft and branch. In the bottom movie, the graft has been placed close to the stenosis, and upstream of the branch, producing much faster and more uniform flow in the host. The implications of this are discussed above.
4.2Mb mpeg
Magnetic resonance imaging (MRI) of a live subject provides the boundary conditions (i.e. geometry and flow rates) for a computational fluid dynamic (CFD) simulation of the blood flow patterns. In this animation, streak length and color intensity are proportional to the blood velocity. Color distinguishes particles exiting the external and internal branches. Note the complex swirling flow patterns and mixing of red and blue particles.
Click here to download a high resolution version of this movie (warning: ~70 Mb)
4.2Mb mpeg
This aneurysm model was reconstructed from an MR angiogram of a dog with an implanted vein-pouch aneurysm. MR data were provived by Dr. Richard Frayne.
1.8Mb animated gif
This series of 32 images (from which the bifurcation model at the top was constructed) was acquired using gated fast spin echo with an in-plane spatial resolution of 313 microns and a slice thickness of 2 mm. Total scan time was under 10 minutes on a 1.5T scanner using a custom RF coil.
296kb mpeg
In this animation, particles shaded blue represent blood flowing to the left, highlighting the region of recirculating blood in the carotid bulb. This is where atherosclerotic plaques commonly occur, suggesting a link between hemodynamics and vascular disease.
304kb mpeg
We observed different patterns of post-stenotic recirculation in models having the same stenosis severity but different stenosis geometries. Since such flow patterns may provide ideal conditions for the formation of the blood clots that cause many strokes, we are currently investigating whether geometric or hemodynamic information can help us better identify patients at risk of stroke.
1.9Mb mpeg
We have developed a 3D segmentation tool based on the idea of an expanding "virtual balloon" that extracts the lumen surface from a series of MR images in under a minute. This is a vast improvement over conventional 2D segmentation and CAD-based serial reconstruction techniques that can require hours even for an experienced operator.
224kb mpeg
By automatically detecting the inner and outer margins of the artery wall using a 2D geometrically deformable model, we can measure wall thickness with a precision 3x greater than that available by manual segmentation.