Abstract #1019

Krooshof P, Postma G, Melssen W, Buydens L

Previous research has shown that clustering of Magnetic Resonance data facilitates the diagnosis of brain tumors. However, the study did not include spatial information in the clustering of Magnetic Resonance Imaging (MRI) data. Additionally, the clustering results were not stable.The aim of this study is to investigate the added value of the inclusion of spatial information in the clustering of multivariate MRI images to facilitate the interpretation of MR data.A recently developed methodology uses spatial information to obtain initial parameters for the clustering of a stack of MRI images. With the clustering procedure, unique segmentations are obtained.