Selecting ICA components is sometimes a complex process that requires viewing IC time series, spectral properties, topography, variance, etc. Every plot also takes time to build, so the interactive approach has a high component of waiting.
Describe your solution
I solved this by taking advantage of the MNE report, and created one with all the components' plots. In order to make the selection process easier, I added JavaScript code that adds buttons and option buttons to select the component to reject, and when done, download a JSON file.
My use case is quite general, although the use of a JSON file needs to be discussed (although it should not be up to the user to remember the numbers!).
The report needs some aesthetic tweaks, but it does it job for now.
Describe the problem
Selecting ICA components is sometimes a complex process that requires viewing IC time series, spectral properties, topography, variance, etc. Every plot also takes time to build, so the interactive approach has a high component of waiting.
Describe your solution
I solved this by taking advantage of the MNE report, and created one with all the components' plots. In order to make the selection process easier, I added JavaScript code that adds buttons and option buttons to select the component to reject, and when done, download a JSON file.
My use case is quite general, although the use of a JSON file needs to be discussed (although it should not be up to the user to remember the numbers!).
The report needs some aesthetic tweaks, but it does it job for now.
Additional context
Code to create report: https://github.com/fraimondo/eeg_cleaner/blob/master/cleaner/report.py Example of a report: https://www.dropbox.com/s/oq9u5heyqrem06v/LS103_LGMM_3%2020170504%201056-icm-lg_mm-egi-epo-ica.html?dl=0