Spot the Artifact
Drill: label the artifact type in real 5-s multichannel segments, get immediate feedback with the tell-tale features highlighted, and keep a running score.
3 claims on this page are unverified. TODO(confirm) marks a specific statement the author has not yet checked against a
primary source. Everything else on this page has been reviewed. Treat a marked claim as
provisional and go to the cited source rather than quoting the sentence.
Modes: drill — the widget below runs in drill. Use Share state to put the exact view in the URL.
What it does
A labelling drill on real EEG. Each item is a 5-second segment with at least 8 channels drawn from a bank of 60 or more labelled segments. You pick a label — blink, saccade, EMG, ECG, drift, pop, line noise, bridging, flat, alpha, drowsiness, or clean — and the widget answers immediately, highlighting the channels and time range that carry the evidence and giving a one-line rationale. A 20-item session reports a score to the lesson’s drill exercise.
Controls
| Control | What it does |
|---|---|
| Label buttons | Choose the artifact type for the current item |
| Hint | Reveal which channels to look at without giving the label |
| Reveal | Show the label, the highlighted region and the rationale |
| Next / session | Advance; a session is 20 items; the score is reported at the end |
| Scale / window | Adjust the µV scale; the window is fixed at 5 s |
What to look for
drill
- Decide on the five axes before you press a button: which channels, what shape, how long, what frequency content, what polarity.
- Frontal-polar, rounded, same polarity both sides: blink. Lateral frontal steps of opposite polarity: saccade.
- A thickened, spiky trace over temporal or occipital sites: EMG. A uniform fuzz at one frequency across channels: line noise.
- One channel with a sharp edge and a slow recovery: pop. One channel drifting off its slot: drift. Two neighbouring channels that are identical: bridging.
- Regular posterior trains at about 10 Hz with nothing else wrong: alpha (not an artifact).
- The rationale after each answer is the lesson; a correct label with a different reason than the rationale is worth a second look.
Used in
- L0.5 The artifact atlas (
drill; the lesson’s checkable exercise is a 20-item session at ≥ 85%).
Data provenance
Items are 5-s derivatives (cropped, channel-subset) of public datasets. Each item records its dataset, subject, run, start time, sampling rate, channels, label, label_source, the detector that proposed the label, and a highlight region; the frame shows the provenance of the current item.
ds-eegbci(160 Hz; no hardware filters; ODC-By 1.0, CC0 on the OpenNeuro mirror): drift, pops, line noise, alpha, clean and muscle items.ds-lemonraw (2500 Hz resampled to 250 Hz; has a VEOG channel; CC BY 4.0 per the descriptor, exact terms TODO(confirm)): blink and saccade items.- ECG and pulse items need a source with an ECG channel: TODO(confirm) (§8). Labels that Phase 0 data cannot supply (for example ECG, bridging) may be absent from the bank;
data/scripts/README.mdlists what is missing. - Every label is
label_source: algorithmic(autoreject/pyprep-style detectors plus the peak-amplitude and spectral rules recorded per item) until the author reviews each item and promotes it toexpert. The widget leavesdraftonly when every item it serves isexpert(§4.5).
Open the code
site/src/components/widgets/w-spot-the-artifact/ — Widget.svelte, index.ts, README.md. Repository link: TODO(confirm) (GitHub org/repo, §13 item 3).