Open jbwexler opened 1 year ago
Looks like inconsistent data. The images are LSP with PhaseEncodingDirection j/j-. When we reorient to LAS, the PhaseEncodingDirection becomes k/k-, which topup can't handle. But they're also labeled dir-ap
/dir-pa
, so something's gone wrong in the curation.
We could have sdcflows detect these cases and drop them rather than get to the point of crashing. Then it would fall back to SyN-SDC.
Holding off on this dataset for now until Paul inspects it.
I ran four subjects and this occurred for one: