Using LCModel & Whole Tissue Representations for the Classification of Single Voxel 1H Spectra of Paediatric Brain Tumours

Felix Raschke¹, Nigel Davies², Martin Wilson², Andrew Peet², Franklyn Howe¹

This study takes forward recently published work using the widespread analysis tool LCModel for the direct classification of single voxel 1H MR spectra of the paediatric brain tumours Medulloblastoma (MDB), Astrocytoma (AG) and Ependymoma (EPD). LCModel is designed to estimate individual metabolite proportions by fitting a linear combination of metabolite spectra to an in vivo MR spectrum, but here is used to fit representations of complete tumour spectra. The best classification according to the highest estimated tumour proportion in a leave-one-out analysis compared well to previously published work with 100.0% and 87.7% for MDB vs. AG and MDB vs. AG vs. EPD respectively.