Using pre-operative CT

[cboulay]

For each of our DBS patients, we acquire preop T1 & T2 about 1 week before surgery, preop CT the day before surgery with fiducial screws in the skull, and postop CT the day after the surgery.

This might be naive, but it seems intuitive to me that the pre-post coregistration might be better using images from the same modality (i.e., pre-CT with post-CT) and the cross-modal coregistration might work better when one of the images doesn’t have the DBS-lead artifacts (i.e., pre-MRI with pre-CT).

In other words, would you expect the resulting coregistration between pre-MRI and post-CT to be better if there was a pre-CT step in between? pre-MRI <–> pre-CT <–> post-CT? Or does adding the extra step contribute more error than is reduced by the within-modality pre-post coregistration?

In the same vein, in our more recent patients we acquired post-op MRI in addition to post-op CT. Is it expected that pre-post coregistration would be better if it was MRI-MRI instead of MRI-CT?

[andreashorn]

Hi Chadwick,

I’m not sure if it would make a difference since we use linear registrations in these steps. My feeling it that rather the risk of failed coregistrations increases if one has multiple cascaded registrations. But I don’t have any data to support this claim.
I think the “brainshift correction” tool in Lead-DBS should remove any pre- to post inaccuracies, especially if using good quality data.

Regarding postop MR vs. postop CT, this is an ongoing debate and I have not decided. I used to prefer MRIs over CTs since one can verify position in native (postop space), sometimes directly identifying the STN or other structures. I guess the advantage of CT is higher resolution re placement of each contact of the electrodes. Although one can visualize them with multiple MRIs, pretty good as well.

Hope this helps?

[cboulay]

Yes, it helps. Thank you.

One of the reasons I was thinking about this is that when I’m checking the result of the normalization, the pre-op T1 aligns with the MNI template very well but the post-op CT does not align with the MNI template well, despite the pre-MRI & post-CT ‘Check Coregistration’ looking very good. I’ve tried a bunch of different normalization algorithms and I haven’t stumbled upon a solution that gives good results for both pre- and post- normalizations.

[Author]

Posts