metadata
annotations_creators:
- machine-generated
language:
- en
license: cc-by-4.0
multilinguality:
- monolingual
pretty_name: OpenMind2D
size_categories:
- 100K<n<1M
source_datasets:
- AnonRes/OpenMind
task_categories:
- image-classification
- image-to-text
- zero-shot-image-classification
task_ids:
- multi-class-image-classification
- image-captioning
- visual-question-answering
tags:
- medical
- neuroimaging
- brain
- mri
- 3d-to-2d
- computer-vision
- healthcare
paperswithcode_id: openmind
dataset_info:
features:
- name: image
dtype: image
- name: orientation
dtype: string
- name: volume_id
dtype: int32
- name: slice_id
dtype: int32
- name: slice_coord
dtype: int32
- name: split
dtype: string
- name: unique_id
dtype: string
- name: modality
dtype: string
- name: image_quality_score
dtype: float32
- name: age
dtype: float32
- name: sex
dtype: string
- name: health_status
dtype: string
- name: manufacturer
dtype: string
- name: magnetic_field_strength
dtype: float32
- name: repetition_time
dtype: float32
- name: echo_time
dtype: float32
- name: width
dtype: int32
- name: height
dtype: int32
- name: format
dtype: string
- name: file_size
dtype: int32
splits:
- name: train
num_bytes: 16787700000
num_examples: 335754
download_size: 11751390000
dataset_size: 16787700000
OpenMind2D: 2D Brain MRI Slices
OpenMind2D is a 2D medical imaging dataset derived from the OpenMind dataset. It contains 335,754 2D slices extracted from 3D brain MRI volumes in three anatomical orientations (axial, sagittal, coronal).
Dataset Statistics
- Total Images: 335,754
- Resolution: 256×256 pixels
- Format: JPEG
- Size: ~11.7 GB
- Splits: Train (70%), Validation (20%), Test (10%)
- Orientations: Axial, sagittal, coronal
- Modalities: T1w, T2w, FLAIR, DWI, and 19+ additional MRI types
Source
This dataset is derived from the OpenMind dataset (Dufumier et al., 2024), which contains 114,000 3D brain MRI volumes from 800 OpenNeuro datasets.
Processing
- Slice extraction from three anatomical orientations
- Isotropic resampling to 1mm³ spacing
- Intensity normalization (1st-99th percentile clipping)
- Resize to 256×256 pixels
- JPEG compression with metadata preservation
Dataset Structure
OpenMind2D/
├── metadata.parquet # Primary metadata
├── train/ # All images
│ ├── 00000001_000.jpg
│ └── ...
└── README.md
Key Metadata Fields
image
: 256×256 JPEG brain MRI sliceorientation
: axial, sagittal, or coronalvolume_id
: Volume identifierunique_id
: Original OpenMind volume IDmodality
: MRI sequence typesplit
: train/validation/testage
: Subject agesex
: Subject sexmanufacturer
: Scanner manufacturer
Usage
from datasets import load_dataset
# Load dataset
dataset = load_dataset("liamchalcroft/OpenMind2D")
train_data = dataset['train']
# Get sample
sample = train_data[0]
image = sample['image']
orientation = sample['orientation']
modality = sample['modality']
# Filter by modality or orientation
t1_data = dataset.filter(lambda x: x['modality'] == 'T1w')
axial_data = dataset.filter(lambda x: x['orientation'] == 'axial')
Citation
If you use this dataset, please cite the original OpenMind work:
@article{dufumier2024openmind,
title = {OpenMind: A Large-Scale Dataset for Self-Supervised Learning in Medical Imaging},
author = {Dufumier, Basile and others},
journal = {arXiv preprint arXiv:2412.17041},
year = {2024},
url = {https://arxiv.org/abs/2412.17041}
}
License
This dataset is released under the Creative Commons Attribution 4.0 International License (CC BY 4.0), consistent with the original OpenMind dataset.