Abstract:
The main objective of the project was to develop an automatic diagnostic tool using
magnetic resonance imaging (MRI), in order to facilitate the medical doctors in their decision making
process, about the existence of brain tumor at its earlier stage. The proposed framework is a step
towards building a clinical decision support system to automate the detection of brain tumor. The tool
worked on magnetic resonance images (MRI), therefore training and testing dataset of images were
obtained from an open- source Harvard medical school online repository. The algorithm comprised of
following five steps, i.e. pre-processing, features extraction, training the classifiers, testing and final
evaluation of framework. Two types of features were extracted after pre-processing the MRIs. The
extracted features were utilized for training the classifiers i.e. support vector machine (SVM), K
nearest neighbor (KNN) and naive bayes. KNN and naive bayes showed 100% accuracy, specificity
and sensitivity, where as SVM was 70% a
