We propose a methodology for discriminating between various types of normal and diseased brain tissue in medical images that utilizes Vector Quantization (VQ), an image compression technique, to extract discriminative texture features. Rather than focusing on images of the entire brain, we direct our attention to extracting local descriptors for individual regions of interest (ROIs) as determined by domain experts. After determining regions of interest, we generate a "locally optimal" codebook representing texture features of each region using the Generalized Lloyd Algorithm. We then utilize the codeword usage frequency of each codeword in the codebook as a discriminative feature vector for the region it represents. Finally, we compare k-nearest neighbor, neural network, support vector machine, and decision tree-based classification approaches using the Histogram Model (HM) distance metric. Combined T1 and T2 classification accuracies in mice averaged 89% under certain experimental settings, indicating that our approach may assist radiologists and surgeons in determining disease margins and tissue homogeneity and support construction of brain atlases and pathology models.