Feature selection for brain tumour classification using ratios of metabolites' peak areas from MRSI data Feature selection for brain tumour classification using ratios of metabolites' peak areas from MRSI data Πλήρης Δημοσίευση σε Συνέδριο Conference Full Paper 2015-10-252008enThe main goal of this study was threefold. Firstly, to investigate the significance of the brain metabolites' peak area Ratios derived from the MRSI spectra provided, by performing brain tumor binary classification and comparing the result with those obtained by using only the metabolites' Peak Amplitudes. Secondly to measure the influence of the four magnetic resonance image intensities (MRI) on the classification process and thirdly to estimate the minimum but also biologically significant set of featureshttp://creativecommons.org/licenses/by/4.0/1-66th European Symposium on Biomedical Engineering ESBME A. Heerschap Zervakis Michalis Ζερβακης Μιχαλης G.J. Postma L.M.C. Buydens M.E. Blazadonakis Basic medical sciences Basic sciences, Medical Biomedical sciences Health sciences Preclinical sciences Sciences, Medical medical sciences basic medical sciences basic sciences medical biomedical sciences health sciences preclinical sciences sciences medical