Hello,

I'm attempting to perform regressions between PET and MRI (ASL, specifically) data using the BPM toolbox. When reading up on the model II regression, I thought that this would be the most appropriate option because there are observational errors associated with both modalities. However, when I change which modality is the dependent variable (e.g. PET as dependent, MRI as independent, and vice versa), I get wildly different results. I'm also using a single non-imaging covariate in the model, but entering the same way for both analyses. 
The residual images I get from both analyses are also very different, although I expected them to be the same based on the publication in Neuroimage by Yang et al. I'm assuming this is not an expected outcome, do you have any ideas what might be happening? I'm happy to provide any files or outputs that might be helpful.

Thanks very much,

Dan