Detection and Classification of Brain Tumours from MRI Images Using VGG16 Deep Learning Algorithm

Abstract

The objective of this paper is developing an accurate and effective system for detecting and classifying brain tumors in MRI images. Brain tumors are considered serious diseases that require early and accurate diagnosis, and MRI images are an essential tool in the diagnosis process. However, manually analyzing these images is time-consuming and subject to human error. Therefore, this work seeks to take advantage of deep learning techniques to develop an automated system that can facilitate the diagnosis process and improve its accuracy. The research relied on the use of the pre-trained VGG16 algorithm, which is one of the most famous algorithms used in the field of computer vision. This algorithm was trained on a large data set of MRI images consisting of 3000 images divided equally between healthy and diseased images. The data was divided into 33% training and testing, training was done for 10 iterations, and the model’s performance was evaluated using a variety of criteria including Accuracy, Recall, and F1-Score, which is the most important evaluation criteria in digital images. The results showed that the model achieved a training accuracy of 99.9% and a testing accuracy of 98.8%. The confusion matrix also showed that the model succeeded in correctly classifying 500 healthy samples and 479 infected samples correctly, in terms of evaluation criteria, and the results were outstanding, as the Precision, Recall, and F1-Score reached about 99% for healthy images and 98-99% for images. Infected. These results indicate that the pre-trained VGG16 model was effective in classifying brain tumors with high accuracy, with balanced performance in classifying both normal and lesioned images.