Hi All,
One of our groups ran into an issue with their dataset where the helium recycler started to check the depth of the liquid helium during an active scan. This happened every 60s for most of the task. There is a giant spike up and down after the helium level sampling. I have a fix where I can stitch the two sides of the jump together, but it leaves some small residual artifact afterwards.
I was hoping to use the oversampling temporal projection for this to remove these jumps, but it seems to have no effect on these artifacts. Since they aren’t true flux jumps or pops, it is possible that the underlying assumptions of the algorithm is violated (there is some ringing - about 50ms from the ADC filter). I have used both duration=10s and 2s, but the algorithm doesn’t seem to have any affect on these artifacts.
Code:
import os, os.path as op
import mne
import numpy as np
topdir = '/fast2/ARTIFACT_testing'
os.chdir(topdir)
fname = op.join(topdir, 'GOPAVXNI_SPiN_20260213_001.ds')
raw = mne.io.read_raw_ctf(fname, preload=True, system_clock='ignore',
clean_names=True)
raw.pick_types(meg=True, misc=False, ref_meg=False)
raw.crop(0, 100)
clean_raw = mne.preprocessing.oversampled_temporal_projection(raw)
clean_raw.plot()
Here is the plot of the “cleaned” data.
MNE version: 1.10.2
OS: Linux Ubuntu 22.04
Thanks for the help,
-Jeff Stout