 |
 |
In our institute we would like to use FSL-FIRST to compute the volumes of several brain structures for a large cohort of patients. The manual of FIRST states that we should use fslstats to compute the volume from the generated nifti files by counting the number of voxels with the label of the structure (e.g. fslstats output_name_all_fast_firstseg -l 16.5 -u 17.5 -V).
This, to me, seems to be an ineffective way, since we are then making rounding errors. Voxels can contain both the hippocampus and another structure, this partial volume effect seems lost if I look at the nifti files.
Wouldn't it be more accurate to compute the volumes directly from the fitted model?
Is there a command to compute the volume of brain structures from the model instead of from the labeled maps?
Thanks,
Bram