Probabilistic Gabor and Markov Random Fields Segmentation of Brain Tumours in MRI Volumes

In this paper, we present a fully automated technique two stage technique for segmenting brain tumours from multispectral human brain magnetic resonance images (MRIs). From the training volumes, we model the brain tumour, oedema and the other healthy brain tissues using their combined space characteristics. Our segmentation technique works on a combination of Bayesian classification of the Gabor decomposition of the brain MRI volumes to produce an initial classification of brain tumours, along with the other classes. We follow our initial classification with a Markov Random Field (MRF) classification of the Bayesian output to resolve local inhomogeneities, and impose a smoothing constraint. Our results show a Dice similarity coefficient of 0.668 for the brain tumours and 0.56 for the oedema.