Brain tumor detection across diverse MR images: An automated triple-module approach integrating reduced fused deep features and machine learning

• Proposes INDEMNIFIER: a novel hybrid model for brain tumor classification. • Employs data augmentation to enhance model robustness. • Combines pre-trained deep models and PCA for efficient feature extraction. • Demonstrates superior performance in handling noisy and diverse brain MRI data. Brain tumors pose a significant threat to human health due to their potential to disrupt normal brain function. Early and accurate detection is crucial for effective treatment. This study proposes a novel triple-module approach for automated brain tumor classification from MRI images. The first module utilizes pre-trained deep learning models (DenseNet121 and ResNet101) to extract informative features. The second module employs Principal Component Analysis (PCA) for dimensionality reduction. Finally, the third module utilizes a Random Forest classifier for tumor classification. The proposed model is evaluated on multiple datasets, demonstrating impressive performance (accuracy is >90 %) on both high-quality and noisy images. The results highlight the potential of this approach for improving brain tumor diagnosis and treatment planning.