Jan Weis1, Patrik Ring2, Tommie Olofsson3, Francisco Ortiz-Nieto1, Johan Wikstrm1
1Dept. of Radiology, Uppsala University Hospital, Uppsala, Sweden; 2Dept. of Radiology, Karolinska Institute, Stockholm, Sweden; 3Dept. of Pathology, Uppsala University Hospital, Uppsala, Sweden
The sets of MR spectra (TE 30 ms) were acquired from normal brain and gliomas grade II, III and IV. LCModels spectral fits were normalized (L2-norm). Mean measured spectra (MS), mean spectra of macromolecules and lipids (ML) in the range 1.40.9 ppm and mean difference spectra (DS = MS - ML - Lac) were computed. Correlation analysis of the patient and mean normalized spectral amplitudes was used for classification of the tumors. It was found advantageous to perform analysis using DS spectra. The shape of the ML spectrum was found to be good marker to discriminate between glioma grades.