Benjamin Segun Aribisala1, Andrew M. Blamire1
1Newcastle Magnetic Resonance Centre, Newcastle University, Newcastle upon Tyne, Tyne and Wear, UK
A significant step in the analysis of imaging data is accurate definition of regions of interest (ROI) within a single tissue type. Analysis is often done by interactively defining ROI on each image under analysis (e.g. T1 or T2). This approach is sensitive to image resolution which introduces partial volume effects biasing the analysis. We propose a fully automatic multi-parametric approach whereby complementary information in multiple images is considered in order to classify each quantitative image into its tissue classes. We apply this method to brain data and demonstrate it is time efficient and largely free of partial volume effect.