Classification of brain tumors based on magnetic resonance spectroscopy

Nowadays, diagnosis and treatment of brain tumors is based on clinical symptoms, radiological appearance, and often histopathology. Magnetic resonance imaging (MRI) is a major noninvasive tool for the anatomical assessment of tumors in the brain. However, several diagnostic questions, such as the type and grade of the tumor, are difficult to address using MRI. The histopathology of a tissue specimen remains the gold standard, despite the associated risks of surgery to obtain a biopsy. In recent years, the use of magnetic resonance spectroscopy (MRS which provides a metabolic profile, has gained a lot of interest for a more detailed and specific noninvasive evaluation of brain tumors. In particular, magnetic resonance spectroscopic imaging (MRSI) is attractive as this may also enable to visualize the heterogeneous spatial extent of tumors, both inside and outside the MRI detectable lesion. As manual, individual, viewing and analysis of the multiple spectral patterns, obtained by an magnetic resonance (MR) spectroscopy exam, is time-consuming and often needs specific spectroscopic expertise, it is not practical in a clinical environment. Widespread use of MR spectroscopy requires specialized processing and evaluation of the data and easy and rapid display of the results as images or maps for routine clinical interpretation of an exam. In this thesis, different approaches have been developed to process and integrate MRI, MRS and MRSI data for differential diagnosis of brain tumors, and to visualize the obtained results in an attractive manner. This thesis identifies problems that can be encountered during this procedure and provides possible solutions. In particular, feature extraction from MR spectra, (multi-class) classification of MRS(I), integration of MRI and MRSI data and the visual representation of the tissue typing results have been studied. The conclusions and developments of this thesis have been established within the context of two European projects of the Sixth Framework Programme for Research and Technological Development. HealthAgents (Agent-based distributed decision support system for brain tumour diagnosis and prognosis, 2006-2008) and eTUMOUR (Web accessible MR decision support system for brain tumour diagnosis and prognosis, incorporating in vivo and ex vivo genomic and metabolomic data, 2004-2009) aim to develop decision support systems to assist clinicians in decision making. Different scientific contributions of this thesis have been implemented in these decision support systems. This can potentially expand the use of MR spectroscopy in clinical practice to support diagnosis and prognosis of brain tumors, and it may allow individually optimized therapy planning.