Author: Tom Ross (165.112.78.---)
Date: 01-02-09 15:11
I have noticed at (at least some) AFNI programs write unscaled data with the scl_slope and scl_inter header attributes set to zero. However, PyNifti returns scaled data as data*slope+intercept. It was not clear from the format specification (at least I did not find) who has the bug. So:
1. Should any program writing unscaled data set the slope to be 1, or
2. Should any program reading data check to see if the slope is non-zero. If so then multiply by the slope.
On the surface, it seems 1 is the better approach, as it is more consistent.
