Tensor Estimation Module

[ashleyrumple]

When using the tensor estimation module, the final version QCed DWI gets labeled as a tensor. The only workaround is to disable the tensor module.

[styner]

works around: run dti estimation separately (with brain masking preceding it), not as a pipeline. The computation of the tensor incorrectly sets the header of the final output

[styner]

adding detail: when running multi-step protocol, such as EDDY followed by tensor estimation, then the final output "*QCed.nrrd/nii.gz" has the header of a tensor and the data of the final QCed DWI, which is incorrect

[styner]

Running it, only with Slice checker + tensor, produces correct results

[styner]

This issue overlaps with an issue of the tensor estimation module with dtiestim for tensory estimation / nrrd writing. When using dtriestim, the DWI input data provided to dtiestim (in OUTPUT/STEP_DTI_Estimate/input.nrrd) has a "none" too many in the space directions (i.e. space directions is 5 dimensional instead of 4). This will need to be corrected first, before we can test whether the above commit resolved the error.