myousif9 / hippograd

GNU General Public License v3.0
0 stars 0 forks source link

Outputed gradients don't appear to correspond to previously observed hippocampal gradients #4

Open myousif9 opened 2 years ago

myousif9 commented 2 years ago

The current iteration of the pipeline runs to completion without any apparent problems though the gradients outputted don't appear to correspond to previously described hippocampal gradients.

First to show the averaged HCP reference gradients used for alignment:

G1: image

G2: image

Then to show the first two gradients from one subject from another dataset:

G1: image

G2: image

@jordandekraker if you had any thoughts or ideas why this might be happening?

jordandekraker commented 2 years ago

I believe this line https://github.com/myousif9/hippograd/blob/4b67049765260fe623b8397198704f82ae5553e3/hippograd/workflow/scripts/compute_gradients.py#L138 should take gm.aligned as its input. Incidentally, you should probably also save the lambda values

jordandekraker commented 2 years ago

Oh also i think in order for the alignment to proceed to need to specify a method (eg 'procrustes'). I'm not sure about that though, it might just default to procrustes