dartbrains/localizer

---
license: cc-by-nc-4.0
task_categories:
- image-classification
tags:
- neuroscience
- neuroimaging
- fmri
- bids
- brain
- nifti
pretty_name: Dartbrains Localizer
size_categories:
- 1K<n<10K
configs:
- config_name: betas
description: "Individual condition beta maps (NIfTI) with subject and condition labels"
default: true
- config_name: betas_stacked
description: "All-condition stacked beta volumes per subject (NIfTI)"
- config_name: events
description: "Task event files (onset, duration, trial_type) per subject"
- config_name: participants
description: "Participant demographics and metadata"
- config_name: fmriprep_bold
description: "Preprocessed BOLD data from fmriprep (NIfTI)"
- config_name: fmriprep_confounds
description: "Motion and physiological confound regressors from fmriprep"
---

# Dartbrains Localizer Dataset

A subset of the [Brainomics/Localizer](https://doi.org/10.25720/1ca1-0sfd) functional MRI dataset, prepared for the [Dartbrains](https://dartbrains.org) neuroimaging course at Dartmouth College.

## Dataset Description

- **Subjects:** 20 (S01-S20)
- **Task:** Functional localizer (auditory/visual stimuli: sentences, computation, motor tasks, checkerboards)
- **Format:** BIDS-compliant with fmriprep derivatives
- **License:** CC-BY-NC-4.0

## Quick Start

### Load beta maps (recommended for most exercises)

```python
from datasets import load_dataset

ds = load_dataset("dartbrains/localizer", "betas")
img = ds[0]["nifti"] # nibabel.Nifti1Image
subject = ds[0]["subject"] # "S01"
condition = ds[0]["condition"] # "audio_computation"
```

### Load event files as a table

```python
ds = load_dataset("dartbrains/localizer", "events")
# Convert to Polars
import polars as pl
df = pl.from_arrow(ds["train"].to_arrow())
```

### Load a single file directly (for nibabel/nltools workflows)

```python
from huggingface_hub import hf_hub_download

path = hf_hub_download(
repo_id="dartbrains/localizer",
filename="derivatives/betas/S01_betas.nii.gz",
repo_type="dataset",
)

# Use with nibabel
import nibabel as nib
img = nib.load(path)

# Use with nltools
from nltools.data import Brain_Data
brain = Brain_Data(path)
```

### Load specific subjects (selective download)

```python
from huggingface_hub import snapshot_download

path = snapshot_download(
repo_id="dartbrains/localizer",
repo_type="dataset",
allow_patterns=["derivatives/fmriprep/sub-S01/**", "sub-S01/**"],
)
```

### Load tabular data with Polars

```python
import polars as pl

events = pl.read_csv(
"hf://datasets/dartbrains/localizer/sub-S01/func/sub-S01_task-localizer_events.tsv",
separator="\t",
)
```

## Dataset Structure

```
dartbrains/localizer/
├── dataset_description.json
├── participants.tsv
├── participants.json
├── task-localizer_bold.json
├── README
├── sub-S01/
│ └── func/
│ └── sub-S01_task-localizer_events.tsv
├── sub-S02/
│ └── ...
├── derivatives/
│ ├── betas/
│ │ ├── S01_betas.nii.gz # all conditions stacked
│ │ ├── S01_beta_audio_computation.nii.gz
│ │ ├── S01_beta_audio_left_hand.nii.gz
│ │ └── ...
│ └── fmriprep/
│ ├── sub-S01/
│ │ ├── anat/ # T1w preprocessed, transforms
│ │ ├── figures/ # QC reports
│ │ └── func/ # preprocessed BOLD, confounds, masks
│ └── ...
```

## Conditions

The localizer task includes the following conditions:
- `audio_computation` / `video_computation`
- `audio_sentence` / `video_sentence`
- `audio_left_hand` / `audio_right_hand`
- `video_left_hand` / `video_right_hand`
- `horizontal_checkerboard` / `vertical_checkerboard`

## Citation

```bibtex
@article{papadopoulos2017brainomics,
title={The Brainomics/Localizer database},
author={Papadopoulos Orfanos, Dimitri and Michel, Vincent and Schwartz, Yannick and Pinel, Philippe and Moreno, Antonio and Le Bihan, Denis and Frouin, Vincent},
journal={NeuroImage},
volume={144},
pages={309--314},
year={2017},
doi={10.1016/j.neuroimage.2015.09.052}
}
```