Brain Tumor Diagnosis Using Convolutional Neural Network on Magnetic Resonance Imaging Data

Abstract

Brain tumors, characterized by abnormal cell growth, are commonly detected by magnetic resonance imaging (MRI). The primary tumor types– meningioma, glioma, and pituitary tumor– exhibit distinct patterns on MRI scans. This work trains a convolutional neural network (CNN) to perform multinomial classification of brain tumors, using more than 7,000 MRI scans from the publicly available Kaggle dataset.