Deciphering Brain Tumor Diagnosis: A CNN-Based Deep Learning Approach

A brain tumor is defined as the rapid growth of abnormal cells in the brain. Magnetic resonance imaging (MRI) is the best way to diagnose these tumors. Using MRI images, doctors can carefully identify and define specific areas of soft tissue growth to determine the presence of brain tumors. Today, brain cancer diagnosis uses machine learning and deep learning algorithms along with advances in artificial intelligence. Using these algorithms has many benefits, such as increasing the brain’s focus, reducing errors, and providing more accurate results. These benefits increase the efficiency of the decision-making process and make it easier for the patient to choose the most appropriate treatment options. This article introduces the use of convolutional neural networks (CNN) to detect brain tumors and evaluate their consequences. The main goal is to use neural networks, especially deep learning techniques, to identify brain tumors. From the results obtained in the training and testing phase, the pre-trained model achieved an accuracy of 0.9664 and a loss of 0.0934. Obtained are indicators for the classification of glioma, meningioma, non-neoplastic and pituitary tumors. Precision, recall, and F1 scores for glioma were 0.9781, 0.9477, and 0.9627, respectively. In the absence of tumor, precision, recall, and F1 scores were 0.9790, 0.9943, and 0.9866, respectively. The precision, sensitivity and harmonic mean values of the pituitary gland are 0.9589, 0.9951 and 0.9767, respectively. After analyzing the dataset, it is clear that convolutional neural networks (CNN) are the most reliable method for predicting the presence of brain tumors.s