MNI to T1 space transformation of brain parcellations using .h5 transformation matrices and ANTS

Hi neurostars!

I am doing my analysis in T1 space (EPI taken to T1 space, using fmriprep output) and now I want to use some ROIs from available brain parcellations. These ROIs are in MNI soI have used ANTS and .h5 matrices to do the backward registration. The only problem now is that the output ROI is in T1 space resolution (1mm) not the EPI in T1 space (2mm).

From what I understand, the EPI to T1 transformation does minimum change of the data and also does not resample to T1 resolution (which is the reason made me reassured I can use T1 space EPI output instead of EPI orig space that is used more traditionally in subject level analysis). So this EPI in T1 space has the same resolution as original EPI (2mm in my case) not the T1 resolution (1mm in my case). So I’m confused why the T1 to MNI matrix (.h5) seems to have 1mm to 2mm resolution transfer in it. If it has been used to take my T1 EPI to MNI, the T1 space EPI should be 2mm so the backward transformation shouldn’t give me a 1mm resolution.

Not sure if I could properly explain and if any of what I said makes any sense!! I can try to explain better if needed. Would very much appreciate any insight or experience.

Thank you!!!