brain-structure.py

import os
import json
import datasets
import logging

logger = logging.getLogger(__name__)

_DESCRIPTION = """
This dataset contains T1-weighted .nii.gz structural MRI scans in a BIDS-like arrangement.
Each scan has an associated JSON sidecar with metadata, including fields such as subject
demographics, scanner information, and a 'split' field indicating train/validation/test.
"""

_CITATION = """
@dataset{Radiata-Brain-Structure,
  author    = {Jesse Brown and Clayton Young},
  title     = {Brain-Structure: A Collection of Processed Structural MRI Scans},
  year      = {2025},
  url          = {https://huggingface.co/datasets/radiata-ai/brain-structure},
  note      = {Version 1.0},
  publisher = {Hugging Face}
}
"""

_HOMEPAGE = "https://huggingface.co/datasets/radiata-ai/brain-structure"
_LICENSE = "ODC-By v1.0"

class BrainStructureConfig(datasets.BuilderConfig):
    """
    Configuration class for the Brain-Structure dataset.
    You can define multiple configurations if needed (e.g. different subsets).
    """
    def __init__(self, **kwargs):
        super().__init__(**kwargs)

class BrainStructure(datasets.GeneratorBasedBuilder):
    """
    A dataset loader for T1 .nii.gz files plus JSON sidecars.
    Each sidecar includes a 'split' field identifying whether the scan
    belongs to the train, validation, or test set.

    Usage Example:
        ds = load_dataset(
            "radiata-ai/brain-structure",
            name="all",
            split="train",
            trust_remote_code=True
        )
    """

    VERSION = datasets.Version("1.0.0")
    BUILDER_CONFIGS = [
        BrainStructureConfig(
            name="all",
            version=VERSION,
            description=(
                "All structural MRI data in a BIDS-like arrangement, labeled "
                "with train/validation/test splits."
            ),
        ),
    ]
    DEFAULT_CONFIG_NAME = "all"

    def _info(self):
        """
        Provides metadata about the dataset, including feature types
        and general dataset information.
        """
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "id": datasets.Value("string"),
                    "nii_filepath": datasets.Value("string"),
                    "metadata": {
                        "split": datasets.Value("string"),
                        "participant_id": datasets.Value("string"),
                        "session_id": datasets.Value("string"),
                        "study": datasets.Value("string"),

                        # Additional fields from the JSON sidecar
                        "age": datasets.Value("int32"),
                        "sex": datasets.Value("string"),
                        "clinical_diagnosis": datasets.Value("string"),
                        "scanner_manufacturer": datasets.Value("string"),
                        "scanner_model": datasets.Value("string"),
                        "field_strength": datasets.Value("string"),
                        "image_quality_rating": datasets.Value("float"),
                        "total_intracranial_volume": datasets.Value("float"),
                        "license": datasets.Value("string"),
                        "website": datasets.Value("string"),
                        "citation": datasets.Value("string"),
                        "t1_file_name": datasets.Value("string"),
                        "radiata_id": datasets.Value("int32"),
                    },
                }
            ),
            homepage=_HOMEPAGE,
            license=_LICENSE,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager: datasets.DownloadManager):
        """
        Returns SplitGenerators for 'train', 'validation', and 'test'.
        Each split is identified by matching the 'split' field in the JSON sidecar.
        """
        data_dir = dl_manager.dataset_dir

        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={"data_dir": data_dir, "desired_split": "train"},
            ),
            datasets.SplitGenerator(
                name=datasets.Split.VALIDATION,
                gen_kwargs={"data_dir": data_dir, "desired_split": "validation"},
            ),
            datasets.SplitGenerator(
                name=datasets.Split.TEST,
                gen_kwargs={"data_dir": data_dir, "desired_split": "test"},
            ),
        ]

    def _generate_examples(self, data_dir, desired_split):
        """
        Recursively scan the data_dir, locate JSON sidecar files, and yield
        examples whose 'split' field matches desired_split.

        Each yielded example includes:
         - 'nii_filepath': pointing to the corresponding .nii.gz file
         - 'metadata': dictionary of subject and scan information
        """
        id_ = 0
        for root, dirs, files in os.walk(data_dir):
            for fname in files:
                if fname.endswith("_scandata.json"):
                    sidecar_path = os.path.join(root, fname)
                    with open(sidecar_path, "r") as f:
                        sidecar = json.load(f)

                    # Only yield if 'split' matches the desired split
                    if sidecar.get("split") == desired_split:
                        # Attempt to locate the matching .nii.gz file
                        # Typically the sidecar is named sub-xxx_ses-xxx_scandata.json
                        # and the NIfTI file: sub-xxx_ses-xxx_T1w.nii.gz
                        possible_nii_prefix = fname.replace("_scandata.json", "_T1w")
                        nii_filepath = None
                        for potential_file in files:
                            if (potential_file.startswith(possible_nii_prefix) 
                                and potential_file.endswith(".nii.gz")):
                                nii_filepath = os.path.join(root, potential_file)
                                break

                        if not nii_filepath:
                            logger.warning(
                                f"No corresponding .nii.gz file found for {sidecar_path}"
                            )
                            continue

                        # Build the example
                        yield id_, {
                            "id": str(id_),
                            "nii_filepath": nii_filepath,
                            "metadata": {
                                "split": sidecar.get("split", ""),
                                "participant_id": sidecar.get("participant_id", ""),
                                "session_id": sidecar.get("session_id", ""),
                                "study": sidecar.get("study", ""),
                                "age": sidecar.get("age", 0),  # default to 0 if missing
                                "sex": sidecar.get("sex", ""),
                                "clinical_diagnosis": sidecar.get("clinical_diagnosis", ""),
                                "scanner_manufacturer": sidecar.get("scanner_manufacturer", ""),
                                "scanner_model": sidecar.get("scanner_model", ""),
                                "field_strength": sidecar.get("field_strength", ""),
                                "image_quality_rating": float(sidecar.get("image_quality_rating", 0.0)),
                                "total_intracranial_volume": float(sidecar.get("total_intracranial_volume", 0.0)),
                                "license": sidecar.get("license", ""),
                                "website": sidecar.get("website", ""),
                                "citation": sidecar.get("citation", ""),
                                "t1_file_name": sidecar.get("t1_file_name", ""),
                                "radiata_id": sidecar.get("radiata_id", 0),
                            },
                        }
                        id_ += 1