Deep learning-based automated brain tumor detection using a proposed convolutional neural network architecture

Abstract

Brain tumors pose a critical threat to human health, often resulting in significant neurological impairments and high mortality rates. Early and accurate detection is essential for improving patient outcomes; however, conventional diagnostic methods, such as magnetic resonance imaging (MRI) interpretation by radiologists, are prone to subjectivity and inconsistency. This study proposes a deep learning (DL)-based solution to automate brain tumor detection using a customized convolutional neural network (CNN) architecture tailored for multi-class classification of glioma, meningioma, pituitary tumors, and non-tumor cases. The model incorporates residual blocks, activation functions, and batch normalization to extract and generalize complex spatial features from MRI images. A dataset of 6,000 labeled MRI images was preprocessed and augmented to enhance robustness. The model was evaluated using standard metrics: accuracy, precision, recall, and F1-score. Comparative analysis against ResNet50, InceptionV3, and MobileNetV2 demonstrated superior performance of the proposed model, achieving a classification accuracy of 99.14% and F1-scores of up to 1.00 across some tumor classes. These results validate the model's potential as a reliable and scalable diagnostic aid, offering improved accuracy and reduced diagnostic time. This work sets the foundation for intelligent, accessible, and clinically deployable brain tumor detection systems.