Velocity-aided Cardiac Segmentation and Physically-based Cardiac Motion Tracking

Motivation

Heart disease is one of the most life-threatening diseases. The accurate diagnosis and appropriate treatment of heart disease are central to the survival of patients. There are numerous diagnostic methods that can assess abnormalities of the heart. Among these methods, cardiac functional analysis has been widely used to derive global and regional parameters that describe functionality of the heart. The accuracy of these parameters depends heavily on cardiac segmentation and motion tracking techniques. Therefore, it is essential to have the accurate and reliable cardiac segmentation and motion tracking techniques.

Velocity-aided Cardiac Segmentation

To derive cardiac functional parameters using various imaging modalities such as ultrasound and MRI, the first step is to extract the myocardial boundaries. Since the myocardial boundaries are assumed to be closed and smooth contours, active contour models have been widely used to extract these boundaries. However, it has been a quite difficult task to find the accurate myocardial boundaries, especially around endocardial boundary, because of the loss of signal due to blood flow or the partial volume effects at the myocardial boundaries and papillary muscles. Furthermore, since active contour models used the final contour from a previously processed temporal neighboring frame as an initial seed for the subsequent frame to be processed, the propagation errors can be significant where there are considerable changes in position and shape of a boundary.

A new segmentation technique that is based on an active contour model and phase contrast MRI is used to address the stated problems. A newly defined external force, called the orientation gradient force, is derived from three-dimensional phase contrast MRI velocity maps. This three-dimensional orientation gradient force is calculated by the tensor based orientation estimation technique and is combined with a generalized active contour model to extract the myocardium boundaries in the initial frame. Then, sequential frame processing using the velocity integration technique is performed to facilitate the automatic segmentation process of sequential MR images.

Physically-based Cardiac Motion Tracking

It has been known for many years that heart wall motion is a sensitive indicator of heart disease such as ischemia. Moreover, clinicians take abnormalities in heart wall motion very seriously because their extent can result in subsequent morbidity and mortality. However, a proper characterization of heart wall motion still remains a challenging problem, because of difficulties in obtaining data correspondence between different time phases and reconstructing the through-plane motion of the hearts.