Good question. First of all, accuracy not only depends on the contrast, but also on the resolution. In fact, I would say that resolution is the most important thing to consider.
Next thing to consider is the coverage you would like to compute shape analysis only. For example, between a 1mm iso and a 0.5x0.5x3mm axial scan, if you only compute shape analysis on a single slice, then the 0.5x0.5x3mm axial resolution is better (because 0.5x0.5 in-plane resolution is more precise than 1x1 mm). However, if you aggregate measurement over a larger superior-inferior distance, then it might be better to choose the 1mm iso resolution for the following reasons:
- if the cord is curved, there will be less partial volume effect with 1mm-thick slices (as opposed to 3mm-thick slices).
- there will be more slices covering the S-I region, hence more statistical power and less sensitivity to a segmentation issue happening at one slice.
Now, in terms of contrast, the T2* being a gradient echo with long TE, it could be more affected by signal dropout (affecting the segmentation)
The T1w MPRAGE is sensitive to participant motion and cord pulsatile effects
The T2w is, in general, cleaner.