Combining Transfer Learning with CNNs and Machine Learning Algorithms for Improved Brain Tumor Classification from MRI

Accurate identification of brain tumors from magnetic resonance imaging (MRI) scans is a necessity in the realm of diagnosis and the ensuing treatment. In this paper, we investigate the benefits of using transfer learning with the convolutional neural network (CNN) architectures EfficientNet and DenseNet to extract relevant features. Next, we compare various machine learning approaches, notably artificial neural networks (ANN), random forest, and support vector machine (SVM), for brain tumor classification. Our experiments were carried out on a dataset comprising MRI images of different classes of brain tumors. The experimental findings showcase the efficacy of transfer learning in feature extraction and reveal differentiated performance between classification algorithms, with a best accuracy of 96.78% for ANN. This study contributes to the advancement of brain tumor classification and provides valuable insights for the choice of machine learning methods in this field.