Incorporate TRX #1006
Incorporate TRX #1006
Conversation
| bundles) | ||
| bundles, | ||
| reference=reference) | ||
|
|
||
| seg_sft.sft.to_rasmm() | ||
| for bundle_name in seg_sft.bundle_names: |
There was a problem hiding this comment.
We should be able to use the fact that the trx has groups here in a more sophisticated manner.
AFQ/tasks/decorators.py
Outdated
| @@ -7,6 +7,10 @@ | |||
| import nibabel as nib | |||
| from dipy.io.streamline import save_tractogram | |||
| from dipy.io.stateful_tractogram import StatefulTractogram | |||
|
|
|||
| from trx.trx_file_memmap import TrxFile | |||
There was a problem hiding this comment.
Imports from trx need to be optional
|
This is now ready for review. It is working on my laptop with the option of running it as usual with trk outputs, or, by adding |
|
What if we made this as optional for storing the results of bundle recognition, but always used for the results of the initial tractogaphy. This would save disk space and I think would have no downsides (we don't load the entire tractographies into MRtrix except on rare occasions, and we can just convert it then). |
…e going to generate a TRX or TRK
LazyTractogram is used here specifically in support of TRX. This will make the documentation/API at the user level more clear.
|
If these pass, this is good to merge |
| elif streamlines.endswith(".trx"): | ||
| is_trx = True | ||
| trx = load_trx(streamlines, img) | ||
| trx.streamlines._data = trx.streamlines._data.astype(np.float32) |
There was a problem hiding this comment.
Is it necessary to cast to float32 here? If the trx is saved at float16 this would double memory usage of this without increasing precision.
There was a problem hiding this comment.
this was so it would play nice with certain dipy cipy functions which I guess assume float32. We will have to make these functions compatible with float16 in the long run, but I figured this was good for now.
This supersedes #818, because the rebase became too complicated.