CNN based Deep Learning Approach for Brain Tumor Detection in MRI Images

Abstract

A brain tumor must be detected through its initial stages, or it can show some serious symptoms that cannot be treated after it has advanced. Most brain tumors are uncovered only after the first symptoms are seen. Ordinary performance of the human body is severely affected by a brain tumor. Medical image processing (MIP) is crucial to identify and label brain tumors. Magnetic Resonance Imaging (MRI) is widely used for identifying and labelling brain tumors. Classification models built on Deep Learning (DL) architectures have gained popularity in recent years for accurately identifying brain tumors in MR images. The intent of this study is to identify, whether the patient is having brain tumor or not using MRI scans. A DL architecture is showcased in this study for accurately identifying brain tumors in MR images. The dataset employed in this study is publicly accessible from Kaggle and comprises of 3000 images, 1500 of which have tumor and 1500 does not. The showcased architecture has produced phenomenal results in identifying brain tumors compared to various other innovative models given in the past. Various performance metrics like F1 score, recall, precision, and accuracy are employed to test the showcased architecture's performance. This showcased architecture was able to attain 98.66% accuracy in identifying brain tumors.