A Scientific Approach to Building an Image Classification model of brain MRI images for Brain Tumor detection

Abstract

The Computer associated-learning is one of the most significant achievements in the field of medical imaging. Generally, we use different technologies like computer tomography (CT scan) to diagnose disease or injury; in the lungs, liver, brain, etc. For this research, we have used MRI image datasets for the identification of Brain tumor classification and non-tumor classification. Tumors are generally abnormal growth; if this type of growth occurs in the brain is called a brain tumor. Early detection and proper treatment may reduce the chances of cancer. Computer vision is the domain where image features and classification extraction can be done very efficiently. In this research, automatic types of MRI image data sets can be considered using CNN (convolution neural network), i.e., VGG 16 architecture. With the help of a pre-trained model classifying and detecting brain tumors of an existing data set can be done. After augmentation volume of the data set can be improved and using individual augmentation values of each class also can be found. Experimental result shows that an accuracy of 97 % of an online data set can be seen.