Non-invasive techniques such asMagnetic Resonance Imaging (MRI) and Magnetic Resonance Spectroscopy (MRS) are often required for the diagnosis of tumours for which conclusive biopsies are not commonly available.While radiologists are used to interpretingMRI, many of them are not accustomed to make sense of the biochemical information provided by MRS. In this situation, oncology radiologists may benefit from the use of computer-based support in their decision making. As part of the AIDTumour research project, the analysis of MRS data corresponding to various tumour pathologies is used to assist expert diagnosis. The high dimensionality of the MR spectra might obscure atypical aspects of the data that would jeopardize their automated classification and, as a result, the process of computerbased diagnostic assistance. In this study, we put forward a method to overcome this potential problem that combines methods of visualization through non-linear dimensionality reduction, automatic outlier detection, and radiologists’ expert opinion.