Brain Tumor Classification Using MRI Images and Deep Learning Methods

A brain tumor is an abnormal mass in the brain that can be benign or malignant depending on the nature of the constituent cells. The origin of the tumor may be from the brain tissue, or it may spread to the brain from another place, or it may metastasize. If not diagnosed in the early stages, it can be life-threatening. Therefore, accurate prognosis of brain tumor in the early stages is very important for its diagnosis and treatment. The traditional methods used to diagnose brain tumors are sampling and examining Magnetic resonance imaging (MRI) or CT scans by humans. However, examining the large number of MRI images by humans that require expertise is tedious and impractical, so there is a basic need to automatically classify a brain tumor image. In this paper, an algorithm for the diagnosis and classification of brain tumors is presented using transfer learning. Basically, instead of building a new model, we used a pre-trained model (ResNet 50) to extract the feature and classification. We have also presented another method based on the combination of transfer learning (TF) and support vector machine (SVM) algorithm. Finally, these two methods were compared and it was found that the use of machine learning algorithms can have a significant effect on classification accuracy.