Enhanced brain tumor classification using convolutional neural networks and ensemble voting classifier for improved diagnostic accuracy

Abstract

Brain tumors, characterized by abnormal cell growth within the brain and surrounding tissues, present significant clinical challenges. Early and accurate detection is essential for effective diagnosis, treatment planning, and improving patient outcomes. Magnetic resonance imaging (MRI) is the preferred modality for brain tumor detection due to its ability to produce high-quality images without ionizing radiation. This study addresses the need for accurate classification by leveraging three pre-trained convolutional neural network models – DenseNet-201, ResNet-101, and SqueezeNet – which enhance feature extraction and classification accuracy. The models were evaluated with and without K-fold cross-validation to ensure robust and reliable results. Additionally, implemented an ensemble voting classifier (EVC) to combine the strengths of the individual convolutional neural network (CNN) models, leading to improved accuracy and robustness. The models were tested on two datasets: (i) a binary dataset and (ii) a multi-class dataset, demonstrating the versatility of the approach. The ensemble classifier achieved 99.69% accuracy for multi-class data and 100% for binary data, outperforming individual models. Key metrics such as accuracy, sensitivity, specificity, precision, and F1-score were used to assess performance. These results highlight the effectiveness of ensemble learning for magnetic resonance imaging brain tumor classification, providing valuable insights for future research and potential clinical applications.