Cardiovascular Magnetic Resonance

We aim to develop new methods and indices to quantify cardiac function using clinical CMR images. We create biophysical models of heart structure and function to allow fast, integrated analysis of time-varying three-dimensional MRI data. Model-based image processing algorithms allow incorporation of high level expert knowledge into the low level tasks of image analysis. The technological advances arising from this work will significantly enhance the clinical evaluation of patients and treatment benefit.

We have developed clinical CMR analysis software, which is being used clinically at Auckland Hospital and clinical research sites worldwide. Our biophysical models will further understanding of the processes underlying heart failure, pressure overload hypertrophy, myocardial infarction and diabetic cardio-myopathy. This research is a multidisciplinary collaboration including cardiologists, radiologists, physiologists, computer scientists and bioengineers.

Three-dimensional display of magentic resonance short and long axis images of the heart, in relation to a mathematical model of the left ventricle. The model is fitted to the images interactively using guide-point modelling (Radiology 2000; 216(2):597-602). Ventricular mass and volume are calculated up to the mitral valve plane (yellow disc), in order to correct for errors due to the longitudinal motion of the base.

A deformable MR phantom was built to test methods of estimating stress and material stiffness from non-invasive MR images. This shows a cyclindrical phantom undergoing inflation. Tagged MR images allow measurement of 3D deformation (strain). A knowledge of the boundary conditions allows calculation of stress (coloured) and material stiffness.

MRI can also be used to study the microstructure of cardiac muscle. This picture shows the orientation of the muscle fibre architecture in a cross-section of a heart. Blue denotes an angle of –60 degrees, green 0 degrees (in the plane of the image) and red 90 degrees.

MRI tagging allows calculation of 3D strain maps in the beating heart. Also, a Gadolinium contrast agent can be injected intravenously to image regions of scar in the heart muscle. This picture shows the 3D outline of a large anterolateral myocardial infarction (red crosses), in relation to the 3D strain map obtained from MR tagging. Blue is normal contraction strain (20%) and red is impaired (0%) contraction.

The Cardiovascular Magnetic Resonance Group gratefully acknowledges the support of its funding partners:

• Health Research Council of New Zealand
• NHF