Detection of Brain Tumor by Mining fMRI Images

Abstract: Brain tumor patients are increasing day-to-day. This paper proposes a novel approach to extract metabolite values from graph. Metabolites like NAA, Creatine, Choline and Cr2 are used to detect the brain tumor. Cho/NAA ratio plays most important role in deciding the tumor type so weights are assigned to each metabolite while clustering. Clustering algorithms could able to achieve accuracy up to 86%. Proposed system is based on decision tree algorithms which are proven to be better against clustering algorithms. Proposed system stores the metabolite values in dataset instead of storing fMRI images so reduces the image processing tasks and memory requirements.

Keywords: fMRI, Clustering algorithms, Classification algorithms, Z-score ranking, K-means, FT, C4.5 algorithm.