N. T. Doan1, S. van Rooden2, M. J. Versluis2, A. G. Webb2, J. van der Grond, M. A. van Buchem, J. H. C. Reiber1, J. Milles1
1LKEB - Department of Radiology, Leiden University Medical Center, Leiden, Zuid Holland, Netherlands; 2CJ Gorter Center - Department of Radiology, Leiden University Medical Center, Leiden, Zuid Holland, Netherlands
The aim of this work is to investigate and compare the information content of T2*-weighted magnitude, phase and SWI data by means of texture analysis. Textural features of the cerebral cortex on magnitude, phase or SWI images were derived based on Gray Level Co-occurrence Matrix. Subsequently, classification using respectively magnitude-based, phase-based and SWI-based features was carried out with a Fisher linear discriminant classifier and a leave-one-out cross-validation approach. The results suggest that, from a texture information perspective, phase images have higher information content than either magnitude or SWI images.