A question for John Ashburner and Ged Ridgeway,
Im unsure how to interpret the voxel values of the output of "Save Jacobian Rate" in your longitudinal registration toolbox.
Can the voxel values be considered as the change in volume (a voxel scaling factor ) between Time 0 and Time 1 in the same way that the Jacobian Determinants in a cross sectional registration can be considered as the change in volume (scaling factor) between atlas and subject?
The description given in the help box: Jacobian rate = J(midpoint to time 1) - J(midpoint to time 0) / interscan interval
won't give rise to a scaling factor.
For example, if
J(midpoint to time 1) = 2, i.e. voxel has doubled
J(midpoint to time 0) = 0.5, i.e. voxel has halved
Then the voxel has increased by 4 between Time 0 and Time 1, but
Jacobian rate = J(midpoint to time 1) - J(midpoint to time 0) = 2 - 0.5 = 1.5 ~= 4
I think J(atlas to time 1) / J(atlas to time 0) would give the scaling factor between Time 0 and Time 1.
Thanks
Ben
|