PCA-based metrics for T1 mapping

[Mobbyjj]

Hi, thank you for providing fantastic implementations and showcases for registration. I tried to use the PCAmetric for T1 mapping from the ModelZoo.
(https://github.com/SuperElastix/ElastixModelZoo/tree/master/models/Par0039/par_groupwise/par_real_data). According to the manuscript (https://doi.org/10.1016/j.media.2015.12.004), T1 mapping should follow the exponential fitting, which I assume some frames should be inverted and arranged according to the inversion time (Images adpoted from https://doi.org/10.1016/j.jcmg.2015.11.005. As in this case, the images are rearranged to Im1 to Im8 according to the inversion time and some magnitude of frames are multipled by -1 to follow the fitting).

However in the showcase you provided, I think they are not rearranged accordingly and all the intensity of the frames are positive, which may not follow the T1 fitting. The images are cropped and reference point is shown from https://github.com/SuperElastix/ElastixModelZoo/tree/master/models/Par0039/par_groupwise/par_real_data

I wonder any pre-processing or normalisation you have used for this showcase. Or whether you can provide the inversion time that I can try for the fitting?

Looking forward to your reply.