Scan table columns

The scan table is one table with one row per recording. The Converter shows it in the Table view; a scan from the command line writes it as a tab-separated file you name (the GUI calls it inventory.tsv), and a project keeps it in .bidsmgr/project/scans/<scan>/inventory.tsv. The two are the same table: the Table view shows a selection of the file's columns, some under shorter names.

Three words mean three things in the scan table, as in BIDS: the modality is how the data was acquired (mri, eeg, meg, pet), the datatype is the folder it lands in (anat, func, eeg), and the suffix is what the recording is (T1w, bold, eeg).

In the Table view

Manage columns... under the table shows, hides and reorders these; the columns marked Always cannot be hidden. Editable columns can be changed in the table, in Properties or with Bulk edit.... The descriptions are the ones Manage columns shows. Where the file names a column differently, its name in the file is written beneath it.

Column	Editable	Shown at first	What it holds
(tick)	Yes	Always	Whether this row is converted. Untick to skip the series.
(state)	No	Always	At-a-glance state badge: ok, warning, error, skipped, non-image, or physio.
subject `participant_id`	No	Always	Participant this row converts under, as sub-XXX.
dataset	No	Yes	BIDS dataset slug. Rows with different datasets become sibling BIDS roots.
session	Yes	Yes	Session this row converts under, as ses-XXX. Blank when the study has no sessions.
modality	No	Yes	How the data was acquired: mri, eeg, meg, pet, nirs. Not the same as the datatype.
datatype	Yes	Yes	The BIDS folder the row lands in: anat, func, dwi, fmap, eeg, meg, pet. An MRI acquisition can land in any of the first four.
suffix `bids_guess_suffix`	Yes	Yes	BIDS suffix (T1w, bold, dwi, physio, ...).
task	Yes	Yes	Task entity (task-XXX) for func / eeg / meg rows.
run	Yes	Yes	Run index (run-N) when a series was repeated.
confidence `bids_guess_confidence`	No	Yes	How sure the classifier is about the datatype and suffix, from 0 to 1.
origin `source_folder`	No	Yes	Where the row came from: the folder inside the raw tree that holds its files.
format	No	Yes	What the row was read from: DICOM, ECAT, FIF, EDF, BrainVision, EEGLAB, CTF.
sequence	No	Yes	Scanner sequence name used for classification. Empty for formats that do not name one.
BIDS name `bids_name`	No	Yes	The BIDS filename this row will be written as, without its extension. Edit the entity columns to change it.
recognised as `sequence_kind`	No	No	What the classifier recognised the sequence as: T1w, bold, dwi, fieldmap, scout. MRI only.
backend	No	No	Converter backend that will handle the row: dcm2niix, mne-bids, ecat, or bidsphysio.
source file `source_file`	No	No	Source recording path (EEG / MEG). Blank for DICOM rows.
n_files	No	No	Number of source files in the series.
acq_time	No	No	Acquisition time from the DICOM / recording header.
image_type	No	No	DICOM ImageType (ORIGINAL / DERIVED, MAGNITUDE / PHASE, ...).
study_date	No	No	Study date, used to cluster longitudinal sessions.
study name `StudyDescription`	No	No	DICOM StudyDescription (the study / protocol name). Read-only, MRI only. Not a BIDS entity; shown for awareness. Multiple distinct values in one scan raise a warning.
PatientID	No	No	DICOM PatientID, a key part of subject identity.
GivenName	No	No	Patient given name from the header.
FamilyName	No	No	Patient family name from the header.
sex `PatientSex`	Yes	No	Participant sex (M/F/O). Seeded from the header; editable; written to participants.tsv.
age `PatientAge`	Yes	No	Participant age in years. Seeded from the header; editable; written to participants.tsv.
hand `Handedness`	Yes	No	Participant handedness (R/L/A). Seeded from the recording header; editable.
n_chan `n_channels`	No	No	EEG / MEG channel count.
sfreq	No	No	Sampling frequency (Hz) of the recording.
duration `duration_sec`	No	No	Recording duration in seconds.
line_freq	Yes	No	Power-line frequency (Hz) written to the sidecar. Dropdown (50 / 60); blank uses the dataset default.
montage	Yes	No	EEG / MEG montage applied on conversion. Dropdown of MNE built-in montages; blank leaves positions as-is.
reference `eeg_reference`	Yes	No	Per-row EEG reference electrode override (e.g. Cz). Blank uses the dataset default.
ground `eeg_ground`	Yes	No	Per-row EEG ground electrode override. Blank uses the dataset default.
companion `companion_files`	Yes	No	Already-curated companion files (events / beh / stim) linked to this row and copied into BIDS on convert. Edit via the Properties panel.
probe_nifti `probe_n_nifti`	No	No	NIfTI files dcm2niix actually produced in a --probe-convert run.
probe_vols `probe_n_volumes`	No	No	Volumes dcm2niix actually produced in a --probe-convert run.
repetition `repetition_type`	No	No	Repeat classification, e.g. suspected_abort (operator restart).
issues	No	No	Scanner-detected notes: non-image series, B0 reroute, missing entity, ...

Only in the file

These columns are in inventory.tsv but not in the table. They are what BIDS Manager uses behind the scenes, and what a script can read.

Column	What it holds
`subject`	The subject as found in the source: for DICOM the given name in the header's patient name, for EEG and MEG the label found in the path. The BIDS label is in `participant_id`.
`entities`	Every entity of the row as JSON, such as `{"subject": "001", "session": "01", "task": "rest"}`. The BIDS name is built from this column: edit it, then run `bidsmgr-rebuild`.
`bids_path`	Where the row goes inside its subject folder: the datatype folder and file name for DICOM rows, the base name for other rows and after `bidsmgr-rebuild`.
`series_uid`, `study_instance_uid`	The DICOM series and study identifiers.
`rep`	The repeat number when the same series was acquired more than once.
`study_time`	The time of the study, from the header.
`bids_guess_classifier`, `bids_guess_datatype`, `bids_guess_entities`, `bids_guess_skip`	Which classifier named the row, and what it proposed. `bids_guess_skip` is true for a row it decided not to convert.
`classification_confidence`, `classification_alternatives`, `classification_evidence`	The parameter classifier's confidence, the other names it considered, and the evidence for its choice. Why this name in Properties shows them.
`fieldmap_targets`	The runs a fieldmap row corrects, when you chose them in Properties.
`probe_n_files`, `probe_extensions`	What a probe conversion wrote for the row (Settings > Scan).
`tracer_suggestion`, `radionuclide_suggestion`, `recon_method_suggestion`, `recon_filter_suggestion`, `injected_dose_suggestion`	PET values read from the header, offered in the dataset metadata and in Properties. A suggestion is never written unless you take it.
`montage_suggestion`, `manufacturer_suggestion`	For EEG and MEG, the montage whose channel names match best, and the manufacturer from the header or the file format. Suggestions too.
`n_times`, `recording_time`, `has_positions`	For EEG and MEG: the number of samples, the start of the recording, and whether the file holds sensor positions.

Columns whose names start with _ are kept for BIDS Manager's own use inside a project and are not written by bidsmgr-scan.

Editing the file

A spreadsheet program can edit inventory.tsv, as long as it saves it as tab-separated text. Change the entities column (the session, the task, the run), the include column (1 or 0), or the datatype and suffix, then run:

Terminal

$ bidsmgr-rebuild inventory.tsv

bidsmgr-convert rebuilds the names itself before converting, so a table you edited and did not rebuild still converts under the names its entities say. Change participant_id and entities together: see Known issues.

An inventory.tsv written by BIDS Manager before 1.4.1 uses older column names and cannot be read; scan the raw data again.