# OpenBrain v1.0 reviewer sample

This directory contains the public reviewer sample for OpenBrain v1.0.
It is intended for quick inspection without downloading the full release.

## Contents

- `sample_manifest.csv`: metadata and QC-7 values for the sample.
- `cases/<case_id>/image.nii.gz`: preprocessed T1-weighted image.
- `cases/<case_id>/brain_mask.nii.gz`: SynthStrip-derived brain mask.
- `cases/<case_id>/whole_brain_segmentation.nii.gz`: released automated
  whole-brain segmentation.

The sample contains 64 source-diverse cases. All sample cases are drawn
from the HighConf tier and were selected by low combined risk `R`, with
Q1/Q3/Q4 visual-cleanliness tie-breaks. The sample is therefore designed
to show high-confidence examples for reviewer preview and smoke testing.

These labels are algorithmic pseudo-labels, not manual ground truth, and
should not be used for clinical decision-making.

## Download

```bash
hf download openbrain-anon/openbrain_v1_0 \
  --repo-type dataset \
  --include "sample/**" \
  --local-dir openbrain_v1_0_sample
```

## Preview

This sample directory includes a standalone preview script:

```bash
pip install nibabel matplotlib numpy
python openbrain_v1_0_sample/sample/preview_sample_case.py \
  --case-dir openbrain_v1_0_sample/sample/cases/<case_id> \
  --output preview.png
```