mne-tools
Feature: Pupil Deconvolution (port from pyeparse)
Reference issue
This PR implements the last feature proposed for my GSoC in #11879 that I didn't get to over the summer.
What does this implement/fix?
This basically ports the pupil deconvolution code from (the unmaintained?) https://github.com/pyeparse/pyeparse , and which was used for @larsoner and @drammock 's pupil deconvolution paper in 2016.
Additional information
I made some minor changes to the code and docstring to work with MNE's API and follow MNE conventions (and make the code more modern i.e. use format specifiers instead of percent specifiers).
To check that the ported code in this branch behaves similarly to the code in pyeparse, I re-ran the scripts to deconvolve the data and generate figure 3 in Dan and Eric's paper (The original scripts used the deconvolution method from pyeparse, and I updated the scripts to use the MNE API in this branch:
| Library | Figure 3 |
|---|---|
| Pyeparse | |
| MNE |
Possible Alternatives
I have not opened an issue for deconvolution specifically, so if there are concerns about merging this into MNE or other ideas, don't hesitate to share.
