Skip to content

Catalog enrichment: per-dataset channel count + electrode-system classification #854

Closed
3 of 3 issues completed

Description

@neuromechanist

Summary

To power a channel-density / electrode-system filter on the Discover page, the D1 catalog needs two new per-dataset fields the website can filter and sort on. Today neither channels nor n_channels appears on any /datasets row (verified across all 646 rows), so the website's "Has 10-20 montage" sidebar control is a dead stub and the plumbed channels range filter has nothing to act on.

Requested fields

  1. n_channels (int) — representative EEG/MEG channel count for the dataset (e.g. modal value across *_channels.tsv, or max). Enables a density filter: low (≤32), standard (33–64), high (65–128), high-density (129+).
  2. electrode_system / montage (string enum) — best-effort classification of the cap/montage, e.g. 10-20, 10-10, 10-05, biosemi, egi-geodesic, other. Derivable from:
    • channel-label patterns (Fp1/Cz/Oz… → 10-xx; E1..E128,Cz → EGI geodesic; A1..B32 → BioSemi),
    • *_electrodes.tsv / *_coordsystem.json presence + EEGCoordinateSystem,
    • CapManufacturer / CapManufacturersModelName in *_eeg.json when present.

Why this matters to users

Researchers care more about "is this high-density?" than "is it exactly 10-20." A density dropdown plus a system dropdown (geodesic/BioSemi/10-xx) lets people find caps comparable to their own. The website filter UI is being built now in a disabled/guarded state and will light up automatically once these fields land.

Website side

  • FilterState already carries a channels: {min,max} range (URL params c_min/c_max); only the UI + data are missing.
  • Discover already pages the full filtered set client-side for range filters (see website listAllDatasets), so no count/pagination work is needed once the field exists.

Activity

  1. added
    P2Normal priority - standard features and improvements
    featureNew feature or enhancement
    backendBackend/API related
    on Jun 28, 2026
  2. added 2 commits that reference this issue on Jun 28, 2026
  3. neuromechanist commented on Jun 29, 2026

    @neuromechanist
    ContributorAuthor

    Delivered + backfilled on prod

    The epic is complete and the production backfill has converged (channel-montage-sweep → remaining: 0):

    • Migrations 0054 (n_channels, electrode_system) + 0055 (channel_montage_checked_at) live (releases v0.8.71 / v0.8.72).
    • Classifier (channel-montage.ts, MNE-grounded label sets) computes both columns during enrichment and via the admin sweep; the BioSemi false-positive found in the dry-run was fixed (require bank-label majority).
    • List + detail endpoints project both columns.
    • Prod distribution: 392 / 575 EEG-matching datasets classified — 10-10: 252, egi-geodesic: 40, 10-20: 31, 10-05: 23, biosemi: 11, other: 35.

    The website channel-density + electrode-system filter consumes these columns.

    The residual unpopulated rows (iEEG predicate over-match, orphaned nm0002xx repo pointers, annexed/sparse channels.tsv) are diagnosed and tracked in #867 — a follow-up, not a blocker for this feature. Closing.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    P2Normal priority - standard features and improvementsbackendBackend/API relatedfeatureNew feature or enhancement

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions