| T O P I C R E V I E W |
| erin.douglas |
Posted - Aug 05 2005 : 10:23:02 AM
FYI, currently data from the 3T scanner is flipped, and does not line up correctly with the anatomical images when processed with bputil. |
| 8 L A T E S T R E P L I E S (Newest First) |
| syam.gadde |
Posted - Nov 19 2005 : 3:17:39 PM
quote:
Originally posted by price
It is indeed old data, and the orientations are off between the anat and func--for certain they're flipped on the y axis, & I can't tell about the x. The runs are in the 303_ _ and 304_ _ range. Thanks.
In that case, it is more likely that both x and y were flipped than just y. If your data was affected, you should change the functional data only (the anatomicals should be correct).
I believe showsrs2 does not do anything to flip the x and y axes based on the orientation vectors -- it just displays them in the order they are stored in the image files, which is why you didn't see any difference after changing the BXH headers. showsrs2 does flip based on the z direction, to make sure that, for example, the top of the head shows up on the top.
However, bputil pre-processing steps will use the orientation vectors to reorient the images into a standard orientation for SPM. Bad orientation vectors will result in bad normalizations, so if your normalizations look good (and the anatomical and functional data match up), then you should be OK. |
| price |
Posted - Nov 18 2005 : 4:37:37 PM
It is indeed old data, and the orientations are off between the anat and func--for certain they're flipped on the y axis, & I can't tell about the x. The runs are in the 303_ _ and 304_ _ range. Thanks. |
| syam.gadde |
Posted - Nov 17 2005 : 09:12:04 AM
quote:
Originally posted by price
I am assuming that I change the direction parameters corresponding to the anatomical BXH only ?
The particular problem you are referring to was resolved in August, to the best of my knowledge. You should not have to edit your BXH headers any more. Dharol's post refers to a recent unrelated problem with 3T data, which is why this topic came up again. If you do notice that functional and anatomical data are coming up with different orientations, or you are using old data that was affected by this problem, please post again. |
| price |
Posted - Nov 14 2005 : 11:30:06 AM
I am assuming that I change the direction parameters corresponding to the anatomical BXH only ?
When I made the changes to the anatomical BXH, then displayed the functional and anatomical in showsrs2 it did not seem to affect the orientation or position that the displyed images. Any thoughts?
|
| tankersley |
Posted - Nov 14 2005 : 09:00:38 AM
My functional data from 3T still has the grey line artifacts, which I've been told were due to z-gradient malfunctions that GE attempted to fix last Wednesday, November 9th.
My funtional data from Exam #30972 on november 11th run on the 3T from 9-10AM still has these problems in the following runs at these TRs:
Run1: TR 64-76
Run2: TR 68-80
Run3: TR 51-76
dharol
|
| jimmy.dias |
Posted - Aug 05 2005 : 11:32:17 AM
All 160 spiral-in functional runs from the 3.0T reconstructed prior to 8:40 a.m. 2005-08-05 have been affected by the bug posted earlier.
Syam's fix above of negating signs for both the x and y dimensions should work. Please correct the BXH files for your functional runs before doing any pre-processing.
If you've done any kind of preprocessing already, you'll need to reprocess this data after making the appropriate fixes or apply fixes to all the preprocessing results. Applying fixes to preprocessing results will only work on datasets pre-normalization. I advise those of you who have run normalization to redo your processing using bputil, making sure you have the overwrite option enabled.
List of Exams Affected:
30208
30241
30250
30262
30298
30299
30385
30386
30394
30411
30415
30427
30433
30438
30450
30457
30458
30470
30472
30473
30474
30495
30496
30504
30505
30506
Jimmy |
| syam.gadde |
Posted - Aug 05 2005 : 10:33:40 AM
We believe the problem only affected spiral-in scans on the 3T, and we also believe this has been resolved for scans starting today. We are trying to find out how many scans this has affected, and then will post details.
It seems that the reconstruction flipped both the x and y dimensions. If you are affected, merely changing the sign of the three values in the <direction> field for both the "x" and "y" dimensions in your BXH files should allow bputil and other processing programs to work correctly. |
| jamie.morris |
Posted - Aug 05 2005 : 10:27:37 AM
Hua, Jimmy, and Syam are working on this issue right now. It should be corrected soon. In the meantime, you can modify the Y dimension of the BXH file and then run it through BPUTIL. |
