Region Of Interest Based Image Classification: A Study in MRI Brain Scan Categorization
Ashraf Elsayed, Frans Coenen, Marta Garc�a-Fi�ana, Vanessa Sluming · InTech eBooks · 2012
This thesis describes research work undertaken in the field of image mining. More specifically, the research work is directed at image classification according to the nature of a particular Region Of Interest (ROI) that appears across a given image set. Four approaches are described in the context of the classification of medical images. The first is founded on the extraction of a ROI signature using the Hough transform, but using a polygonal approximation of the ROI boundary. The second approach is founded on a weighted subgraph mining technique whereby the ROI is represented using a quad-tree structure which allows the application of a weighted subgraph mining technique to identify feature vectors representing these ROIs; these can then be used as the foundation with which to build a classifier. The third uses an efficient mechanism for determining Zernike moments as a feature extractor, which are then translated into feature vectors to which a classification process can be applied. The fourth is founded on a time series analysis technique whereby the ROI is represented as a pseudo time series which can then be used as the foundation for a Case Based Reasoner. The presented evaluation is directed at MRI brain scan data where the classification is focused on the corpus callosum, a distinctive ROI in such data. For evaluation purposes three scenarios are considered: distinguishing between musicians and non-musicians, left handedness and right handedness, and epilepsy patient screening.