Data archive

Generated 2026-11-11 14:01  ·  scanned X / W / Z  ·  dataset level = second-level directory of each volume

Filter by modality and research focus

The outer ring is modality, the inner ring research focus. The rings always show the whole-group distribution so the relative size of categories stays comparable; categories excluded by the current filter are dimmed but remain clickable (clicking one widens the scope). The right-hand panel lists the matching datasets — click a card and the table below scrolls to that row.

Dataset details

A further 0 directories hold code / tools / documents and are excluded from search and statistics

Needs attention

Grouped by volume / group / project, and nothing below that level — no file paths are listed. One misconfigured permission on a directory tree produces dozens of records with a single root cause; grouping keeps the signal from drowning, and the group named here is where the fix belongs.

How modality and focus are inferred

Neither dimension lives in the registry — both are parsed from directory names. The group already uses a project_modality_date_version convention (e.g. sca3_cj_fmri_20261109) and the parser only accepts words that are actually present; anything else is recorded as Unlabelled, not guessed. When the name offers no clue, the second-level subdirectory is used as secondary evidence (e.g. a subdirectory containing rfmri means fMRI) and the evidence word is written to the modality_ev column of facets.csv so it can be checked. To override a call by hand, edit the modality / focus columns there and set manual to 1; classify.py will not overwrite it.

Other limits: the dataset level is the second-level directory of each volume; W and Z hold very large numbers of small DICOM/EEG files, so no full stat pass was run and this page deliberately shows no capacity comparison (only 16 datasets have a measured size). Owners are merged from owners.csv; missing entries are shown as not registered rather than dropped.