Enhancement of Pre-Trained Deep Learning Models to Improve Brain Tumor Classification

Abstract

Cancer of brain tumor is considered one of the most dangerous diseases in human life, and its early detection may help specialists in treating it. Unless diagnosed early, brain tumors cause a shortening of life and are very destructive. The classification of brain tumors is the important process of determining the type of tumor present in a patient's brain based on medical imaging (MRI) and other diagnostic tests. This process is important for providing an accurate diagnosis and determining the best course of treatment. Artificial intelligence science provides useful and helpful techniques in detecting and classifying brain tumors, based on deep learning and computer vision. This research focuses on providing a better methodology for brain tumor classification, the proposed model was implemented using five pre-trained models: CNN, ResNet101, InceptionV3, VGG16, and VGG19, and enhancement their performance using data augmentation, the achieved Precision, Recall, and F1-Measure of unseen images were 95%, 95%, 95%, 97%, 95% respectively, and tested using three open datasets. In addition to improving the early detection of tumors, these accuracy improvements lead to fewer disabilities, such as paralysis. Data augmentation is a good way to improve the performance of the model by rotation, scaling, and flipping of the images in the dataset. This helps the model to generate new images with better quality