Dear Elisabeth,
I don't know whether this is the reason for your error message, but as you state it's MP2RAGE images: Have you already suppressed the background noise via some masking or does your image forwarded into segmentation look like this http://w1.siemens.ch/home/ch/de/healthcare/produkte/acit/projects/pages/advanced-mp2rage.aspx ? In the latter, try to create a brain mask based on intensity-thresholded TI2 images and apply this mask onto the corresponding MP2RAGE images. Settings in segmentation algorithms are usually optimized for T1-weighted like the MPRAGE, but the MP2RAGE has this bright background noise.
You might also be interested in the "robust" images calculated from MP2RAGE data, see O'Brien et al. (2014, PLOS ONE, https://dx.doi.org/10.1371/journal.pone.0099676 ). Less uniform, but also less noise.
Best
Helmut
|