Skip to content

OpenNeuro (on*) mirrors miss electrodes.tsv/coordsystem, breaking the scalp topomap #855

Closed
nemarOrg/website
#136

Description

@neuromechanist

Summary

The EEG viewer's scalp topomap renders only when it can resolve ≥3 electrode positions: measured coordinates the converter embeds from electrodes.tsv/coordsystem.json, or, failing that, channel labels that match a standard 10-20/10-10/10-05 montage. EGI geodesic labels (E1..E128) match no standard montage, so an HBN recording with no measured positions gets no topomap (correct, by design — better than a wrong head map).

Two copies of the same HBN data diverge in exactly this file set:

dataset electrodes.tsv coordsystem.json channels.tsv viewer topomap
nm000103 (HBN-EEG NC) 3512 3512 3522 ✅ renders (129-ch EGI)
on005516 (HBN Rel. 11, OpenNeuro mirror) 0 0 3397 ❌ TOPO disabled
curl -s https://data.nemar.org/nm000103/v2.0.0/manifest.json | grep -c electrodes.tsv   # 3512
curl -s https://data.nemar.org/on005516/v1.0.0/manifest.json | grep -c electrodes.tsv   # 0

The user uploaded both from the same equipment; the OpenNeuro mirror import dropped (or never carried) the per-recording electrode positions. Same HBN data, but the mirror can't show a topo.

Asks

  • During on* OpenNeuro import, preserve *_electrodes.tsv + *_coordsystem.json when the source has them.
  • Where the OpenNeuro source genuinely lacks them but the cap is a known geodesic/standard net (e.g. GSN-HydroCel-129), consider attaching a canonical montage at import so the topomap (and any montage-based QA) can render.
  • This is the file-content sibling of the now-closed OpenNeuro import path doesn't backfill modalities, tasks, subject_count into the D1 catalog #512 (which covered catalog fields, not BIDS file fidelity).

No website change required; the viewer already does the right thing with whatever positions the dataset ships.

Activity

  1. added
    bugSomething isn't working
    P2Normal priority - standard features and improvements
    backendBackend/API related
    on Jun 28, 2026
  2. neuromechanist commented on Jun 28, 2026

    @neuromechanist
    ContributorAuthor

    Forensic update — the import is faithful; source lacks the files

    Checked the OpenNeuro source directly (2026-06-28):

    GET s3.amazonaws.com/openneuro.org/ds005516/sub-NDARAB678VYW/eeg/sub-..._channels.tsv     -> 200
    GET .../sub-..._electrodes.tsv                                                            -> 404
    GET .../sub-..._coordsystem.json                                                          -> 404
    

    So ds005516 on OpenNeuro ships no electrodes.tsv / coordsystem.json — only channels.tsv. The NEMAR on005516 import faithfully mirrored that (electrodes.tsv: 0, channels.tsv: 3397). This is not an import-drop bug; there is nothing to preserve. The NEMAR-curated nm000103 has electrodes.tsv because it was added during curation, not because the OpenNeuro copy had them.

    Revised fix: viewer montage fallback (website), not import synthesis

    HBN uses the EGI GSN-HydroCel-129 net (labels E1..E128 + Cz), which doesn't resolve against the viewer's vendored STANDARD_1005 set, so no topomap. The right fix mirrors the existing 10-20 fallback: extend the website viewer's standardMontageFor / STANDARD_1005 mechanism to recognize EGI (and BioSemi) named-net labels and project their canonical positions, flagged estimated: true (same caveat the 10-20 fallback already uses). This is additive, mutates no scientific data, and fixes the whole class of named-net datasets that ship no electrodes.tsv — not just on005516.

    Synthesizing electrodes.tsv into the dataset at import (the original framing) is the heavier, more opinionated alternative and is not preferred.

    Action: re-scope to the website viewer (src/lib/eeg-viewer/standard-montage.ts). Keeping this issue open as the tracking item; will link the website PR.

  3. neuromechanist commented on Jun 28, 2026

    @neuromechanist
    ContributorAuthor

    Implemented as the viewer named-net fallback in website PR nemarOrg/website#136 (EGI GSN-HydroCel-129/257 + BioSemi 128/256 canonical positions, flagged estimated). Validated against real HBN E1..E128+Cz labels.

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 relatedbugSomething isn't working

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions