Multi-Sequence MRI Dataset for AI-Driven Brain Tumor Analysis Applications

This dataset contains clinically acquired brain MRI slices collected from medical centers in the Sulaymaniyah region of Kurdistan, Iraq, including Hiwa Hospital and Asia International Hospital. It has been developed for the purpose of binary classification of brain MRI scans into tumor and non-tumor categories and is suitable for training deep learning models in medical image analysis. The dataset consists of 542 original MRI slices obtained from real patient cases, including:
• 305 tumor MRI slices from 50 patients diagnosed with brain tumors
• 237 non-tumor MRI slices from 40 individuals without tumors
To increase the dataset’s diversity and robustness for machine learning applications, a range of image augmentation techniques was applied. As a result, the dataset was expanded to a total of 6,606 images, comprising:
• 3,953 tumor images derived from the original 305 slices
• 2,653 non-tumor images derived from the original 237 slices
Each image is saved in standard formats (e.g., JPG) with a consistent resolution of 320×320 pixels. The dataset includes images from several standard MRI sequences, namely: T1-weighted (T1W), T2-weighted (T2W), FLAIR, T1-weighted contrast-enhanced (T1W-CE), ADC, and DWI.This dataset has been carefully curated, reviewed, and annotated by experienced medical professionals, ensuring high accuracy and reliability for use in research and educational settings. It offers a valuable resource for researchers working on medical image classification, particularly in the detection of brain tumors using MRI data.