Hi-
I see that is, indeed, an oversight in being defined in the program's help-- I will add that.
It is the output of including "-debug_briks" in 3dDWItoDT: calculating chi-square results, for goodness-of-(tensor)-fit measures. So, yes, it can be useful for QCing.
Here is the detailed description of its brick values, from 3dDWItoDT's help:
-debug_briks = add sub-briks with Ed (error functional), Ed0 (orig.
error), number of steps to convergence and I0 (modeled B0
volume).
[May, 2017] This also now calculates two goodness-of-fit
measures and outputs a new PREFIX_CHI* dset that has two
briks:
brik [0]: chi^2_p,
brik [1]: chi^2_c.
These values are essentially calculated according to
Papadakis et al. (2003, JMRI), Eqs. 4 and 3,
respectively (in chi^2_c, the sigma value is the
variance of measured DWIs *per voxel*). Note for both
chi* values, only DWI signal values are used in the
calculation (i.e., where b>THR; by default,
THR=0.01, which can be changed using '-bmax_ref ...').
In general, chi^2_p values seem to be <<1, consistent
with Papadakis et al.'s Fig. 4; the chi^2_c values are
are also pretty consistent with the same fig and seem to
be best viewed with the upper limit being roughly =Ndwi
or =Ndwi-7 (with the latter being the given degrees
of freedom value by Papadakis et al.)
--pt