brain_deer.infrastructure.bids.import_convert¶

Convert DICOM or flat NIfTI folders into a minimal BIDS layout.

DICOM import is pure Python (pydicom + nibabel) — no MATLAB, no Lead-DBS binaries. The default session split mirrors the common DBS clinical layout:

ses-preop — structural MR (T1w / T2w / FLAIR / SWI / …)
ses-postop — post-op CT (and CT variants via acq-)
One NIfTI per (session, suffix, acq) — no runaway run-* spam
When several series collide on the same key, keep the latest study date

Classes¶

ImportReviewRow

Editable mapping row shown in the review UI before commit.

Functions¶

`guess_suffix_datatype`(→ tuple[str, str])	Return `(suffix, datatype_folder)` from filename / DICOM / folder hints.
`classify_dicom_series`(→ SeriesPlan \| None)	Map one DICOM series to BIDS session/suffix/acq, or `None` to skip.
`discover_dicom_subject_folders`(→ list[tuple[str, ...)	Find `(subject_id, dicom_folder)` pairs under a cohort or single-patient tree.
`plan_dicom_import`(→ list[ImportReviewRow])	Dry-run: discover subjects/series and return editable review rows (no writes).
`commit_dicom_import`(→ pathlib.Path)	Write reviewed rows to `rawdata/` + Lead-DBS-style `rawimages.json`.
`import_dicom_to_bids`(→ pathlib.Path)	Convert DICOM series into a BIDS tree. Returns dataset root.
`bidsify_flat_folder`(→ pathlib.Path)	Organize loose NIfTI/TRK files into a BIDS-like tree.

Module Contents¶

brain_deer.infrastructure.bids.import_convert.guess_suffix_datatype(filename: str, *, modality: str = '', description: str = '', path_hint: str = '') → tuple[str, str]¶: Return (suffix, datatype_folder) from filename / DICOM / folder hints.

brain_deer.infrastructure.bids.import_convert.classify_dicom_series(series, *, session_mode: str = 'clinical', session_override: str | None = None) → SeriesPlan | None¶: Map one DICOM series to BIDS session/suffix/acq, or None to skip.

brain_deer.infrastructure.bids.import_convert.discover_dicom_subject_folders(root: str | pathlib.Path) → list[tuple[str, pathlib.Path]]¶

Find (subject_id, dicom_folder) pairs under a cohort or single-patient tree.

Cohort (e.g. HOTDBS1): each child dir is one patient → one subject.
Single patient folder: returns that folder alone (not one subject per date).
Flat DICOM dump: returns (folder_name, root).

class brain_deer.infrastructure.bids.import_convert.ImportReviewRow¶: Editable mapping row shown in the review UI before commit.

brain_deer.infrastructure.bids.import_convert.plan_dicom_import(dicom_dir: str | pathlib.Path, *, subject: str = 'auto', session: str = 'auto', series_indices: list[int] | None = None, session_mode: str = 'clinical') → list[ImportReviewRow]¶: Dry-run: discover subjects/series and return editable review rows (no writes).

brain_deer.infrastructure.bids.import_convert.commit_dicom_import(rows: list[ImportReviewRow], output_root: str | pathlib.Path, *, dataset_name: str = 'BrainDeer DICOM import', source: str | pathlib.Path | None = None, progress_callback=None) → pathlib.Path¶: Write reviewed rows to rawdata/ + Lead-DBS-style rawimages.json.

brain_deer.infrastructure.bids.import_convert.import_dicom_to_bids(dicom_dir: str | pathlib.Path, output_root: str | pathlib.Path, *, subject: str = 'auto', session: str = 'auto', dataset_name: str = 'BrainDeer DICOM import', series_indices: list[int] | None = None, session_mode: str = 'clinical', progress_callback=None, review_rows: list[ImportReviewRow] | None = None) → pathlib.Path¶

Convert DICOM series into a BIDS tree. Returns dataset root.

Pure Python — does not call MATLAB or Lead-DBS.

Pass review_rows from the mapping UI to skip re-planning; otherwise plans automatically (no interactive review — use the dialog for that).

brain_deer.infrastructure.bids.import_convert.bidsify_flat_folder(source_dir: str | pathlib.Path, output_root: str | pathlib.Path, *, subject: str = '01', session: str = '01', dataset_name: str = 'BrainDeer BIDSify import', copy_files: bool = True) → pathlib.Path¶: Organize loose NIfTI/TRK files into a BIDS-like tree.