Contributing to MNE-HFO

MNE-HFO is an open-source software project developed through community effort. You are very welcome to participate in the development by reporting bugs, writing documentation, or contributing code.

Development takes place on the collaborative platform GitHub at github.com/mne-tools/mne-hfo.

Bug reports

Use the GitHub issue tracker to report bugs.

Contributing code or documentation

Please see our contributing guide to find out how to get started.

Contributing your own sliding window-based detector

1. First step is opening up a Github issue to describe the new HFO detector.

2. Secondly, one should follow our API for all Detectors, namely the mne_hfo.base.Detector class. The basic hyperparameters of any Detector are:

• threshold: The threshold point used in computation of HFOs. This could be a threshold on a certain metric, like line

  length, if the LineLengthDetector is used.

• window size: The size of the windows in terms of samples

• overlap: The amount of overlap between consecutive windows

One must also implement their own _compute_hfo function, which returns an HFO event array, consisting of an array of (n_chs, n_windows) that is a binary existence of HFO detected or not. It is within this function that all HFO detection algorithm should be implemented.

3. Finally, if your detector has extra hyperparameters then what may be inherited from Detector, then one can just add these to your own __init__() function.

Once you have implemented your detector, then it is always good to provide unit tests for determining correct functionality of the HFO algorithm. This can be done using simulated HFOs using mne_hfo.simulation module, or using real data that you provide.

Hilbert Detector