search

guruucsd / lateralized-components

Submission to OHBM 2016 on functional lateralization using the neurovault dataset.
0 stars 2 forks source link

Bad components #30

Closed atsuch closed 7 years ago

atsuch commented 8 years ago

I was working with 20 components, and found that there a few (1 for R, another for L) components with a very few voxels with extreme values, resulting in the overall dark image. This one example has the max value of 0.43, when other images have max values around 0.02. Any idea why we get this, @bcipolli ?? wb_r_6

bcipolli commented 8 years ago

Never seen it, but glad to take a look tomorrow morning!

atsuch commented 8 years ago

Hmm... in our code, plotting components individually (no comparison between components) when doing ICA for the first time didn't have the colorbar. So I tried adding the colorbar and reran ICA with --force and n-components =5, and got similarly bad components (one each for R, L, and wb: see below ). I might have to rerun QC to see if any new images are corrupting the analysis..

l_component_1 r_component_4 wb_component_3

atsuch commented 8 years ago

@bcipolli, do you see this type of components if you rerun the ICA? Could you help me troubleshoot?

bcipolli commented 8 years ago

I see the first image that you show. I see that the scale bars are strangely high.

bcipolli commented 8 years ago

@atsuch I didn't see any bad components in n=5 as you did. I saw one with an improper scalebar, but looks like it would be OK if rescaled.

bcipolli commented 7 years ago

@atsuch have you seen this since then? I have not... perhaps our qc.py improved enough that these went away?

bcipolli commented 7 years ago

This was fixed along the way; closing.