Improving brain tumor classification using convolutional neural networks (CNNs) For MRI images of the brain

Abstract

Brain tumors are serious and fatal diseases, which often lead to a decrease in human life expectancy. Early and accurate detection of the nature and classification of these tumors is crucial for developing an appropriate treatment plan that can lead to prolonging the lives of patients with these tumors. Manual diagnosis of large quantities of brain MRI images is considered extremely difficult and complex, and requires the diagnosing physician to have high experience and great accuracy in classifying each type of these tumors according to its shape, dimensions, and location in the human brain. Therefore, an intelligent model based on deep learning (DL) must be developed to accurately diagnose and classify brain tumors. In this study, we will propose a new deep learning (DL) model based on convolutional neural networks (CNNs). This model uses a number of algorithms for the initial processing of MRI images and to be input to the convolutional neural network that In turn, it contains, in addition to the input layer, a number of hidden internal convolutional layers, which perform a number of mathematical operations on the input data to extract features from brain MRI images, and a number of pooling layers, which select the most important features from the total of extracted features, and full connection layers, which create paths. Additional neurons between layers. This allows the network to learn complex relationships between features and make high-level predictions. The results obtained by applying the new model achieved an accuracy rate of 99.6% on an MRI brain tumor dataset obtained from a database Kaggle Brain tumor dataset and over relatively small time frames make this model extremely useful for neurologists to help make quick and accurate diagnostic decisions.