Brain Tumor Detection Using Deep Convolutional Neural Network

Abstract

Abstract: This study presents a deep learning method for classifying brain tumors from magnetic resonance imaging (MRI) data is presented. We used a publically accessible dataset that included pictures of meningiomas, gliomas, pituitary tumors, and healthy brains to train a model using the ResNet-18 architecture. To improve model resilience, data augmentation methods such as rotation, color jitter, affine transformations, random horizontal flipping, and scaling were used. The model's ability to differentiate between various forms of brain tumors was demonstrated by its 98.74% training accuracy and 98.70% validation accuracy across 20 epochs. These findings imply that the suggested approach may be a useful instrument for helping medical professionals diagnose brain tumors accurately and quickly.