STI tensor element χ33 looks like a local field map instead of a QSM-like map. Any idea why?

Hello everyone,

I was hoping someone with experience in susceptibility imaging, especially susceptibility tensor imaging (STI), might be able to help me understand what is happening in my data.

We acquired multi-orientation GRE data with 6 orientations using the following parameters: TE1 / deltaTE / TE6 = 3 / 4 / 23 ms, with TR = 27 ms, and flip angle of 12 degrees.

When I reconstruct QSM for each individual orientation, the maps look good. However, the problem appears when I try to estimate the susceptibility tensor elements.

My processing pipeline is the following:

  • ROMEO for phase unwrapping

  • VSHARP for background field removal

  • Coregistration of all local field maps to the straight-head position

After that, I use an asymmetric STI method to estimate the tensor.

The issue is that when I visualize the χ33 tensor element, I would normally expect something that resembles a QSM map. Instead, what I get looks much more like a local field map.

So I was wondering whether anyone has seen this before, or has any idea what could cause it. Could this be related to:

  • a data acquisition issue,

  • an error in preprocessing or coregistration,

  • a problem with the STI fitting/model itself,

  • or something else entirely?

I attached a screenshot showing how the χ33 element looks.

Any suggestions would be greatly appreciated.

Thank you.