Segmentation and Quantification Techniques for Fitting Computer Vision Models to Cardiac MR, CT, X-Ray and PET Image Data

Abstract

The field of medical imaging has experienced an explosive growth in recent years (1990-99) due to several imaging modalities, such as X-ray, computer tomography (CT), magnetic resonance imaging (MRI), positron emission tomography (PET) and spectral positron emission computer tomography (SPECT) (see Stytz et al. [135] for an extensive survey). The digital revolution and the processing power of computers combined with these modalities have helped humans understand to some extent the complex anatomy of the heart and its behavior. There are still, however, some unresolved problems which are linked to computer vision-pattern recognition (CVPR) and clinical cardiology research. The importance of cardiovascular research has increased, due to the inter-linking of effects arising from non-cardiovascular diseases. In the United States alone, the budget for cardiovascular research was $269 billion in 1997. This points towards the national concern and the degree of importance of cardiovascular research. As reported by the American Heart Association (AHA) [136] and the Herald Newspaper, UK [137], heart disease claims an enormous number of lives.

This chapter is based on material previously published in Pattern Analysis and Applications [139] [550].