Hi Paul,

(1) Yes - I have. I assumed it might be related to this, but if I try generating this in a terminal on a different screen, it still comes up on the main mac monitor, and then ends up distorting if shifted away. If needs be, I'll use the mac monitor with fsleyes, I just don't recall this cropping up pre-update?

(2) Perfect, thank you :)

Ah yes, I think this rings some bells from when I was looking into pymigp for SGE job submissions last year. I'm planning on doing dual regression next, I was just QC'ing the components first.

Thanks for your time Paul, have a great week!

-Ryan

###################################

I have not seen this artifact before. By any chance, have you moved the FSLeyes window from your laptop display to an external (non-retina) display after opening FSLeyes? If so, this is a bug which will be fixed in the next FSLeyes version.

(2) there is no power spectra shown for my independent components from my group ICA

This is another bug which will be fixed in the next FSLeyes version. In the meantime, you can fix it by opening the power spectra overlay list (Settings -> power spectra -> overlay list), and toggling visibility for the melodic_IC image.

How does one load fsleyes -ad -s melodic melodic_IC.nii.gz with a background image instead of the mean (via CLI)?

Instead of using the -ad option, you could explicitly specify your background image of choice, and other display settings, e.g.:

fsleyes -smelodic <your_background_image> melodic_IC -un -cm red-yellow -nc blue-lightblue -dr 3 10

How does one interpret the timeseries and power spectra at the group ICA level

If you have used the MIGP option when running MELODIC (this is likely, as it is enabled by default), the time series / power spectra that are saved out are uninterpretable. MIGP is an optimisation strategy which reduces the amount of RAM required for running a group ICA, but has the unfortunate side-effect that the ICA unmixing matrix will not contain the time courses of the identified components.

Ludovica's paper, and the FSLeyes melodic mode, are intended for use at the single-subject level for de-noising/preprocessing, rather than for use with the results of a group ICA. If you would like to look at the time courses that are associated with the components of a group ICA, the current recommendation is to generate per-subject time courses using dual regression.

Paul



On Mon, 22 Aug 2022 at 15:53, Ryan Stanyard <[log in to unmask]> wrote:
Hi Paul/Matt/FMRIB team,

I've updated to the new fsl version today (on macOS Big Sur) and noticed two issues thus far:
(1) image cut-offs are making the gui unusable for checking slices from any view
(2) there is no power spectra shown for my independent components from my group ICA

Image: https://drive.google.com/file/d/107gUAuCXvFkBmbcDJvYtInqcpsuZ3kWI/view?usp=sharing

If I may also query, related to the latter:
+ How does one load fsleyes -ad -s melodic melodic_IC.nii.gz with a background image instead of the mean (via CLI)?
+ How does one interpret the timeseries and power spectra at the group ICA level - I've heard this isn't reliable (with no expansion on this statement) and have also seen Paul's recent post in which he says the power spectra are read from the melodicFT file and not converted directly into frequencies. I'm trying to match the online guidance and the Griffanti paper to ensure my components are plausible, so it helps if I can trust the frequencies. Could you provide a little clarification or guidance with respect to both the timeseries and power spectra displays when in MELODIC mode please (especially if this has shifted across versions)?

Thanks,

Ryan

########################################################################

########################################################################

To unsubscribe from the FSL list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=FSL&A=1

This message was issued to members of www.jiscmail.ac.uk/FSL, a mailing list hosted by www.jiscmail.ac.uk, terms & conditions are available at https://www.jiscmail.ac.uk/policyandsecurity/