Dear All,
With help from lab members, I have put together a pipeline for
pre-processing intracranial EEG (a.k.a. ECoG) using SPM routines and
format, as well as FieldTrip, FASST and in-house tools. This pipeline
was created to ease the specific pre-processing for this type of data by
allowing:
- automatically identifying bad channels based on the variance/ number
of 'jumps' on each channel when compared to all other channels.
- Response-Time epoching with onset baseline correction in raw signal.
- Multiple visualization tools for time series, including
Skewness-Kurtosis and Power Spectrum plots.
- Baseline correction and TF rescale based on the whole time series,
discarding marked artefacts.
- Univariate analysis performed channel by channel, averaging over time
points, with FDR correction.
I needed those functionalities for my research and am sharing in case it
is of any use to someone else.
The pipeline allows for pre-processing of event-related brain activity,
as well as continuous signals (e.g. rest). The code can be found here:
https://github.com/LBCN-Stanford/Preprocessing_pipeline and has a tutorial.
Important note: There is no such thing as a pre-processing pipeline for
ECoG! Each recording is unique and adapting the code is necessary. Most
routines call SPM batches, that can easily be modified for your
application or have flexible inputs.
Happy to answer questions about it.
Best,
Jessica Schrouff
|