Estimation of Arterial Wall Strain Based on IVUS Image Registration
Atherosclerotic plaque rupture is responsible for the majority of acute
coronary syndromes and myocardial infarctions. It seems reasonable that
measurements of plaque mechanical response can be used to assess the
likelihood of plaque rupture. We are evaluating intravascular ultrasound
(IVUS) imaging, a clinically available technique providing real time
cross-sectional images of coronary arteries, for this purpose.
In this study, we have developed a strain estimation method based on
image registration of a pair of IVUS images acquired at the same vessel
site while under different luminal pressure. This 2D processing method
has the ability to overcome in-plane movement of the IVUS catheter and
heterogeneous tissue deformation, therefore increasing the accuracy of
strain estimation. With the 2D strain tensors, both radial and
circumferential strain distributions can be obtained, and color-coded
for display.
An experimental system based on a clinical IVUS system has been designed
and built to acquire IVUS images for validation and evaluation purposes.
Two sets of test image data have been used to evaluate the performance
of the image registration technique. The first set consists of synthetic
target images generated by applying a known displacement field, based
on a thick-walled deformation model, to an IVUS image of a porcine
carotid artery, see Figs. 1 and 2. The second set consists of IVUS
images acquired in a homogeneous, concentric tissue-equivalent phantom
made of 10% polyvinyl alcohol (PVA) cryogel, whose elastic modulus is
controllable and measurable, see Fig. 3.
We also tested the robustness of the algorithm against image noise. We
acquired two IVUS images at the same vessel site and under the same
luminal pressure, a few seconds apart. Then we registered the second
image to a deformed version of the first image using a known
displacement field, see Fig. 4.
We have applied this method to clinical images acquired during a
standard IVUS procedure. A pair of images at the same vessel site,
obtained at different points in the cardiac cycle, is identified on the
basis of cardiac phase retrospectively retrieved from the IVUS sequence.
These experiments have demonstrated the ability of our image analysis
method to estimate the local strain tensor in the arterial wall, and has
the potential to be applied clinically without necessitating new
hardware or a change in surgical procedure.
