Taylor Salo

icon indicating a website link tsalo.github.io icon indicating an email [email protected]

icon company University of Pennsylvania icon location Philadelphia, PA icon location I am a postdoctoral fellow at UPenn in the @PennLINC lab. I am interested in performing reproducible, transparent research with fMRI.

Results 899 comments of Taylor Salo

Improve synthetic with + in task label for `beh/` data

The example is failing in the expected way, which is great!

Issue a warning if there fieldmaps but they were not applied to any sequence

I think a warning is reasonable, but FWIW I've seen a lot of datasets with unused field maps (e.g., multi-echo field maps where only the first echo's data are used,...

Update bokeh requirement from <=3.4.1,>=1.0.0 to <=3.4.2,>=3.4.2

What version of Python are you trying to install the new version on? I think they dropped support for 3.8, which would explain why `makeenv_38` is failing. This PR does...

fmriprep bids output not fully covered by bids-validator

It would probably be best to split off the stuff that fMRIPrep is missing (e.g., `SpatialReference` is missing from some files' sidecars) into an fMRIPrep issue, so that this issue...

Failed to generate the boilerplate

Can you share the full logs?

Can't run TFCE=True in non_parametric_inference

@YukunQu I tried reproducing the problem with a small mask (125 voxels) and a large number of iterations (10000) using the data and code in https://nilearn.github.io/stable/auto_examples/07_advanced/plot_localizer_mass_univariate_methods.html, but I couldn't produce...

Small brains in the "Surface reconstruction" image in the html report

We could use `nilearn.image.crop_img` right before creating the figure to get rid of the empty space. We recently started doing that in QSIPrep.

Only mean centering PCA components

>One thing to watch out for is image voxels that are consistently zero-filled. I think the adaptive masking step should remove any voxels that are all zeroes, so we should...

using --dummy-scans fails when used together with --external regressors

I just came here to open this exact issue! I will work on fixing ASAP.

Change n_iter to max_iter in TSNE parameters

fMRIPrep builds on 3.10 so we won't want to increase our minimum Python version above that. I don't know what the situation is with AFNI.