| Medical imaging modalities, including Computed Tomography (CT) Magnetic Resonance Imaging (MRI) and Ultrasound (US) are critical for the diagnosis and progress monitoring of many cardiac conditions, planning, visualization and delivery of therapy via minimally invasive intervention procedures, as well as for teaching, training and simulation applications.;Image segmentation is a processing technique that allows the user to extract the necessary information from an image dataset, in the form of a surface model of the region of interest from the anatomy. A wide variety of segmentation techniques have been developed and implemented for cardiac MR images. Despite their complexity and performance, many of them are intended for specific image datasets or are too specific to be employed for segmenting classical clinical quality Magnetic Resonance (MR) images.;Graph Cut based segmentation algorithms have been shown to work well in regards to medical image segmentation. In addition, they are computationally efficient, which scales well to real time applications. While the basic graph cuts algorithms use lower-order statistics, combining this segmentation approach with atlas-based methods may help improve segmentation accuracy at a lower computational cost.;The proposed technique will be tested at each step during the development by assessing the segmentation results against the available ground truth segmentation. Several metrics will be used to quantify the performance of the proposed technique, including computational performance, segmentation accuracy and fidelity assessed via the Sorensen-Dice Coefficient (DSC), Mean Absolute Distance (MAD) and Hausdorff Distance (HD) metrics. |
