Thank you so much for the speedy help Francois!
There is one issue though, that it would take me quite some time to redo the analysis. Is there any workaround to this issue (to get the right labels without recomputing the atlases and the source results?