Classification of Brain MRI Images Using Convolution Neural Network and Transfer Learning

Magnetic Resonance Image (MRI) is a non intrusive procedure for obtaining high resolution images of living tissues. Varieties of tissue contrast images that MRI modalities produce provide important structural information, enabling diagnosis and segmentation of tumors. In this paper we come up with an integrated model of Convolutional Neural Network and Transfer learning for automated binary classification of brain MRI images. We use a sixteen layer pretrained network, to distinguish normal and abnormal images. Our experiments suggest good accuracy in classification on test data, and thus will turn out to be useful for efficient automated diagnosis.