Multi-Class Brain Tumor MRI Dataset: Glioma, Healthy Brain, Meningioma, and Pituitary Macroadenoma
Description
This dataset presents a high-quality collection of brain MRI images categorized into four clinically relevant classes: Glioma, Meningioma, Pituitary Macroadenoma, and Healthy brain scans. Each category contains 300 images, resulting in a balanced dataset of 1,200 images suitable for robust machine learning, deep learning, and radiological analysis. All MRI images have been collected from verified medical sources and anonymized to ensure ethical compliance. A verification certificate is included to validate the authenticity and approved use of the dataset. The images capture diverse tumor characteristics, sizes, and anatomical variations, providing a comprehensive resource for academic and clinical research. This dataset is intended to support a range of applications, including but not limited to: brain tumor classification, tumor detection, segmentation research, radiomics analysis, and AI-driven medical imaging model development. Its balanced class distribution and clear labeling make it particularly useful for training supervised learning models.
Files
Steps to reproduce
The dataset was created by collecting MRI scans from verified clinical sources following standard radiological protocols for brain imaging. All scans were originally acquired using hospital-grade MRI systems operating with conventional T1-weighted and T2-weighted imaging sequences routinely used for tumor detection. The data collection process adhered to ethical and medical imaging guidelines, ensuring that all patient information was fully anonymized prior to inclusion. After obtaining the raw scans, images were reviewed and categorized into four diagnostic classes—Glioma, Meningioma, Pituitary Macroadenoma, and Healthy—based on medical reports and radiologist confirmation provided by the healthcare institution. Each MRI slice was pre-screened to remove low-quality, duplicate, or non-diagnostic images. The images were then standardized for research use through preprocessing steps, including format conversion, resolution normalization, and quality verification. The entire workflow was documented, and a verification certificate issued by the data provider is included to confirm the authenticity and authorized use of the dataset. This methodological outline enables other researchers to understand the data origin and follow similar steps if they wish to reproduce the dataset creation or conduct related medical imaging research.
Institutions
- Daffodil International University