Brain volume measurements should both be accurate and precise. Accuracy of brain segmentation techniques is well studied. With the availability of test-retest datasets, precision (low coefficient of variation (COV)) can be investigated. In this work, we studied the COV of the FSL, SPM, and CAT12 software packages on 120 3T brain MR images of three subjects (40 images each) and compare it to previous results of FreeSurfer on this dataset. CAT12 performs best on total gray matter, white matter, and brain volume; whereas FSL has the lowest COV for CSF. COV values should be considered when studying brain volume change.
This abstract and the presentation materials are available to members only; a login is required.