Participate in MUnitQuest!
Both the Data Challenge and the Algorithm Challenge are hosted on Codabench. To participate, register on Codabench, then navigate to the Data challenge and Algorithm challenge!
Data Challenge
Submission types
To obtain a diverse data collection for MUnitQuest that balances realism and label quality, we invite submissions of both experimental and simulated data. In short, depending on the type of data you want to contribute, a submission consists of the following parts:
| Data type | Submission requirements | Additional considerations |
|---|---|---|
| Experimental HDsEMG |
1. Raw EMG signals 2. Labelled motor unit spike trains 3. Force and/or kinematics data 4. Metadata and provenance |
Labelling approach (2-page PDF) |
| Experimental HDsEMG + concurrent iEMG | Raw iEMG signals must be provided; Labelling approach (2-page PDF) | |
| Synthetic (end-to-end, or hybrid) | Model summary (2-page PDF) |
How to prepare your submission
Step 1 — Prepare your EMG data in EMG-BIDS format Data must be submitted in the standardized EMG-BIDS format. Our dataset submission walkthrough walks through the full process, and the web-based metadata tool generates all BIDS sidecar files from five simple CSV files. We also provide a [Python tutorial] Coming Soon.
Step 2 — Prepare your labeled spike trains
For each recording, provide a *_events.tsv file (BIDS-events file) containing the labeled motor unit spike trains. See the example below for the required format. The submission walkthrough covers how to assemble these into your BIDS dataset.
Step 3 — Write a 2-page PDF Include a short description of your labelling approach (for experimental data) or simulation model (for synthetic data). For concurrent iEMG submissions, describe how the invasive reference was used. This PDF will be placed at the root of your final submission (see Step 5).
Step 4 — Upload a metadata-only zip to Codabench
Prepare a .zip of your BIDS dataset that excludes all raw data files (.edf) and upload it to the Data Challenge on Codabench. You can use the following shell command from the parent directory of your BIDS dataset folder:
zip -r submission_metadata.zip <your_bids_folder>/ --exclude "*.edf"
After processing, you will be able to download an HTML report of your submission — review it to confirm everything is in order before proceeding.
Step 5 — Upload the full dataset
If your Step 4 submission is successful, a link will be provided to upload the complete dataset (including the raw .edf files) together with your 2-page PDF placed at the root of the submission.
Algorithm Challenge
Algorithm submission
This is a prediction submission competition. During both the Familiarization Phase and the Showdown Phase, you will be asked to upload, for each recording, a tabular file (recordingName_events.tsv) containing your predicted motor unit spikes (BIDS-events file) together with a log file (recordingName_log.json) describing essential process metadata (further details to be announced). Submissions apply to both tasks (Isometric and Dynamic) independently. To be eligible for awards, you need to share your code openly (e.g., on GitHub) upon the completion of the competition.
Example: how to report motor unit spike trains
Here is a minimal example of the format (BIDS-event file) used for submitting motor unit spike trains (both for labels and algorithm predictions):
| onset | duration | sample | unit_id | description |
|---|---|---|---|---|
| 0.001 | 0 | 1 | 0 | motor-unit-spike |
| 0.005 | 0 | 5 | 1 | motor-unit-spike |
| 0.011 | 0 | 11 | 0 | motor-unit-spike |
| 0.012 | 0 | 12 | 2 | motor-unit-spike |
| 0.016 | 0 | 16 | 1 | motor-unit-spike |
| … | … | … | … | … |
- onset: Onset (in seconds) of the event, measured from the beginning of the acquisition.
- duration: Duration of the event (measured from onset) in seconds. As a motor unit spike can be regarded as a Dirac impulse, its duration is zero.
- sample: Sample index of the event onset (zero-indexing).
- unit_id: Unique identifier (integer value) of the motor unit corresponding to the detected spike.
- description: Human-readable free-text description of the event.