Image Based Biomarkers from Magnetic Resonance Modalities: Blending Multiple Modalities, Dimensions and Scales.

The successful analysis and processing of medical imaging data is a multidisciplinary work that requires the application and combination of knowledge from diverse fields, such as medical engineering, medicine, computer science and pattern classification. Imaging biomarkers are biologic features detectable by imaging modalities and their use offer the prospect of more efficient clinical studies and improvement in both diagnosis and therapy assessment. The use of Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI) and its application to the diagnosis and therapy has been extensively validated, nevertheless the issue of an appropriate or optimal processing of data that helps to extract relevant biomarkers to highlight the difference between heterogeneous tissue still remains. Together with DCE-MRI, the data extracted from Diffusion MRI (DWI-MR and DTI-MR) represents a promising and complementary tool. This project initially proposes the exploration of diverse techniques and methodologies for the characterization of tissue, following an analysis and classification of voxel-level time-intensity curves from DCE-MRI data mainly through the exploration of dissimilarity based representations and models. We will explore metrics and representations to correlate the multidimensional data acquired through diverse imaging modalities, a work which starts with the appropriate elastic registration methodology between DCE-MRI and DWI- MR on the breast and its corresponding validation. It has been shown that the combination of multi-modal MRI images improve the discrimination of diseased tissue. However the fusion of dissimilar imaging data for classification and segmentation purposes is not a trivial task, there is an inherent difference in information domains, dimensionality and scales. This work also proposes a multi-view consensus clustering methodology for the integration of multi-modal MR images into a unified segmentation of tumoral lesions for heterogeneity assessment. Using a variety of metrics and distance functions this multi-view imaging approach calculates multiple vectorial dissimilarity-spaces for each one of the MRI modalities and makes use of the concepts behind cluster ensembles to combine a set of base unsupervised segmentations into an unified partition of the voxel-based data. The methodology is specially designed for combining DCE-MRI and DTI-MR, for which a manifold learning step is implemented in order to account for the geometric constrains of the high dimensional diffusion information.