Brain Tumor MRI Image Dataset

Abstract

Brain tumors pose a significant challenge in medical diagnostics, necessitating advanced computational approaches for accurate detection and classification. This dataset comprises a curated collection of Magnetic Resonance Imaging (MRI) scans categorized into four distinct classes: No Tumor, Glioma Tumor, Meningioma Tumor, and Pituitary Tumor. To ensure data integrity and reliability, an extensive preprocessing pipeline was implemented, including duplicate image removal using perceptual hashing and correction of mislabeled samples through expert verification. All images were standardized to a resolution of 224×224 pixels to optimize computational efficiency. Furthermore, data augmentation techniques, such as salt-and-pepper noise addition, histogram equalization, rotation, brightness adjustment, and horizontal/vertical flipping, were employed to enhance dataset diversity and robustness. This dataset serves as a valuable benchmark for developing and evaluating deep learning models in medical imaging applications, including automated tumor classification, radiomics analysis, and treatment planning. Researchers are encouraged to utilize this dataset for advancing machine learning-driven medical diagnostics, ensuring appropriate validation through collaboration with domain experts.