MRI Analysis Using Deep Learning for the Identification of Brain Tumors

Abstract

Using Convolutional Neural Network (CNN) algorithms and Magnetic Resonance Imaging (MRI) data, our study offers a comprehensive brain tumor prediction system. We investigate the effectiveness of different pre-trained CNN models, such as VGG19, ResNet101, EfficientNetB0, MobileNetV2, InceptionV3, and DenseNet121, in precisely predicting the presence of brain tumors by utilizing cutting-edge deep learning techniques. After extensive testing, the best-performing model is EfficientNetB0, which achieves an astounding accuracy of 97.96%. We also assess the performance indicators of various model architectures and look into their effects. Our results highlight the crucial part CNNs play in automating the identification of brain tumors, improving the effectiveness and precision of diagnosis. Using the TensorFlow and Keras frameworks, we create a scalable and reliable system that can process large-scale MRI datasets. Our work highlights the potential of deep learning to enhance healthcare diagnostics and advances the field of medical image analysis.