Brain Tumor MRI Images with Four Diagnostic Categories

Holding nearly 1400 JPEG images of brain MRIs, split into training and validation sets. The images are labeled into four categories: glioma tumor, meningioma tumor, pituitary tumor, and no tumor. The dataset is intended for classification tasks.

Use Cases

• Train a convolutional neural network to classify MRI images into the four diagnostic categories (glioma_tumor, meningioma_tumor, pituitary_tumor, no_tumor).
• Benchmark image classification models using the provided training and validation splits, which contain 826 and 100 glioma tumor images respectively.
• Analyze class distribution across the training split, which has 395 images labeled as no_tumor compared to over 800 for each tumor type.
• Perform data augmentation on the training set to address potential class imbalance before model training.

Strengths

• Contains nearly 1400 labeled medical images for model development.
• Provides a clear training/validation split for immediate use in classification tasks.
• Includes four distinct diagnostic categories relevant to brain tumor detection.

Limitations

• The dataset is relatively small for deep learning, with under 1500 total images.
• Class distribution is imbalanced, with the 'no_tumor' class having significantly fewer training images (395) than tumor classes (over 800 each).
• Limited metadata is available; specific acquisition parameters or patient demographics are not provided.

Provenance

Source

Author 'benschill' on Hugging Face.

Collection Method

Collection of brain MRI JPEG images, grouped and labeled for classification.

Freshness

Last updated in July 2022.

Geography

Region tag suggests 'us', but specific geographic coverage is unknown.