Detection of Brain Tumor using Convolution Neural Network algorithm

Brain tumors, especially in their aggressive forms, significantly reduce life expectancy. This study aims to advance the early identification and classification of brain tumor by utilizing a comprehensive approach encompassing binary classification tumor or no tumor. Employing advanced techniques such as deep learning algorithms, including Convolutional Neural Networks (CNNs) with transfer learning, employing pre-trained models such as the Visual Geometry Group 16-layer network (VGG16) and the 50-layer Residual Network (ResNet50), custom CNN model 8 -layer network. The appearance of deep learning technologies has markedly advanced the field of medical diagnostics through enhanced medical imaging capabilities, particularly in identifying and diagnosing various diseases. Within this context, the CNNs have evolved as the leading machine learning algorithm for tasks involving visual learning and image recognition. Our study presents a novel CNN framework that is specifically tailored to classify brain cancers using T1-weighted contrast enhanced MRI into 2 categories: tumor, or no tumor. The results show that our CNN model Achieved (97%) accuracy which is approaching one of the best models VGG-16 (99%), while others score about ResNet-50 (91%), still our model have a good accuracy with minimum computational overhead which is advantages. Keyword (Brain Tumor, Medical Imaging, Deep Learning, Convolutional Neural Networks (CNNs), VGG16, ResNet50, Custom CNN Model, MRI (Magnetic Resonance Imaging)).