ABSTRACT
In this chapter, the authors provide the description of the development of an alternative automated method to extract the contours of the left ventricle in blood pool Single Photon Emission Computed Tomography (SPECT) images. The method combines wavelets theory, neural networks, and fuzzy logic for the extraction of contours for use in computing the Left Ventricle Ejection Fraction (LVEF). The authors describe how the kinds of images they work with are obtained and how the information they encode has to be interpreted. The authors present the overall strategy of the method we suggest for processing the image sets to determine the LVEF. These are a wavelets based pre-processing of the data, a segmentation into regions of interest using neural network approaches and a recognition technique built around fuzzy logic components. The authors also present representative results obtained using fuzzy logic method.
