3D reconstruction bug VE11E

[fedgiove]

Hi, I just installed the R017 pre-realease 6 (build 25 Apr) on a Prisma VE11E, and I tested the new features.

There is some severe bug in the 3D EPI if multiband is enabled. The sequence stops at the end of the first TR, after prescan. I guess it locks at the end of the first dummy scan or possibly at the end of the single band reference (i did not check timing with much attention).

Exam component of the host is freezed (see attachment), and the scanner is left in an inconsistent state: restarting host and mars is not sufficient, a full reboot is needed.

3D without MB looks to work.

2D works as previously, except a slight increase of minimum TR for SE single band (used for fieldmapping), compared to R016. Is that expected?

I'm ready to do any test can be useful to solve the issue.

Thank you, Federico

[eauerbach]

Is that freezing reproducible if the system is rebooted first? Sometimes that happens randomly, but usually it is not reproducible and can't be traced to a particular sequence.

I wouldn't expect TE to change between R016 and R017 on the same system. Can you send details of the protocol?

[fedgiove]

Hi, freezing is completely reproducible, also immediately after reboot (and a localizer scan). It's triggered by just setting a MB factor >1 on an otherwise working 3D protocol.

min TE did not change; what changed is minimum TR. I noticed it because I had some spin echo fieldmap scans set exactly at TR= 7 s, that was roighly the minimum. They have been re-set to an higher minimum value (something like 7.037 s) after upgrade. It's not a problem in itself, of course.

I'll send details about the protocols next Monday. exar or pdf?

Thank you,

[eauerbach]

I can't replicate the 3D+MB issue, at least starting from a default protocol. So if you can send me an example protocol for that, in addition to your TR changing protocol, I will look at it. I can't really do anything with a exar file; a pro file would be ideal (can be generated by dragging a protocol into a windows explorer window), otherwise pdf.

[fedgiove]

Hello, Please find in attachment an example protocol that crashed. However, on my scanner the issue can be easily reproduced starting from any 2D protocol, including the default protocol.

After a little bit of investigation I found the issue happens on many 3D protocols, including non MB, but in MB ones it's systematic. Any other sequence (including your 2D EPI) works as expected

You can find:
Here a movie showing how to trigger a sequence crash starting from the default protocol
Here an instance where the sequence actually crashed the scanner (a forced reboot was required)
Here an example of a 3D protocol that works with MB=1, but crashes with MB=2

Best, Federico
T2_w_3DEPI_1.3mm_ISO_WIP_CMRR.zip

(PS you have to accept the security exception to download, movies are in a local server with an unsigned certificate)

[fedgiove]

I have found an instance when the reconstruction crashes also in 2D, and it's when online analysis is enabled.

EDIT After just more testing I realized that R2016 crashes as well if online processing is enabled and phase images are saved - this is understandeable and probably unrelated

BR, Federico