search

NILOIDE / CMR_intensity_3Dmoco

We propose a method to mitigate the effect of inter-slice motion in CMR images for all SA and LA slices simultaneously by optimizing the 3D rotation and translation parameters on sampled intensities along slice intersections.
2 stars 0 forks source link

Reproducing the problem on UKbiobank data #2

Open Breeze-Zero opened 3 weeks ago

Breeze-Zero commented 3 weeks ago

Hello @vasl12 @NILOIDE , thank you very much for your work. I am using your code to test the data of UKbiobank. Although I do not know how you preprocess the dcm of UKbiobank, I saved the dcm respectively in la2-4.nii.gz, sa.nii.gz according to your code rules, and then ran to this point. Since the code comments are not clear, I had a bit of trouble parsing the issue after this point, so I came to you for help image

NILOIDE commented 3 weeks ago

Hi @Breeze-Zero, Did you figure out what the issue was? I'm suspecting maybe the dimensions of v1 or v2 are incorrect? The trace says something about "Batch element 0". So maybe the data you passed in had a slice with weird dimensions?

Breeze-Zero commented 3 weeks ago

Hi @NILOIDE I should have found the code that you preprocessed into nii, and optimize_affines.py can be run correctly after batch processing according to that code. However, my previous input was preprocessed by myself, and there seems to be some mismatch. Since I don't need the time dimension at the moment, I will try to make some fine-tuning of the code in the future. Thank you for your timely reply.

Breeze-Zero commented 3 weeks ago

Hi @NILOIDE, I'm sorry I seem to have found another problem. How do I merge the corrected SA slices back into the full volume for analysis, since the affine matrix of each layer has changed and it seems impossible to simply stack the arrays.

Breeze-Zero commented 1 day ago

Hi @NILOIDE, I think I need help. I have been trying to solve this problem for some time, but it has not been successful, especially the overlapping parts.😭

Hi @NILOIDE, I'm sorry I seem to have found another problem. How do I merge the corrected SA slices back into the full volume for analysis, since the affine matrix of each layer has changed and it seems impossible to simply stack the arrays.