Open Lestropie opened 5 years ago
I experience the same issue as mentioned in the forum thread and would like to share my observations hoping that it may help in resolving the bug:
[PROBLEM] The segmentation image is re-oriented (as compared to the input T1 image) after "5ttgen fsl" despite the use of the "-nocrop" flag.
[EXPECTATION] The segmentation image should have the same dimension and orientation as the original T1 image. It should be possible to load both images in, for example, FSL view (segmentation image & T1 image) and overlay them.
I would support the suspicion that the reason for this behavior is the use of the fsl tool "standard_space_roi". In my case, I have noticed that the re-orientation only occurs, when the input MR image is oriented in a non-standard way. For example: (A) no re-orientation happens, i.e. the segmentation image overlays perfectly fine the T1 image in FSLView when the T1 image is oriented the following way: Orientation: Axial X axis orientation: left to right Y axis orientation: posterior to anterior Z axis orientation: inferior to superior
(B) re-orientation happens, i.e. the segmentation image DOES NOT overlay the T1 image (FSLView prints: "Unable to load incompatible overlay!") when the T1 image is oriented like Orientation: Sagittal X axis orientation: anterior to posterior Y axis orientation: inferior to superior Z axis orientation: left to righ
OK, that does actually give me a useful place to start when trying to reproduce. Thanks for the input @BennyKay :+1:
From forum thread.
What might be happening is that an image downstream of the
standard_space_roi
step is being used as the output image template, but that step is itself cropping the image FoV independently of the latermrcrop
step.