Samseg Aseg Test | March 17 2017

Andrew ran samseg on the Buckner40 test set. mri_compute_seg_overlap was used to compare the subject's segSubSpace.mgz with its aseg.mgz.


The mean overlap of the majority of subjects was above 0.8, but 2 subjects had a mean below 0.5 and 1 subject failed samseg with the error "maximalDeformation is too small"

fig 1: mean dice scores for all Buckner40 subjects

The low dice scores seen in 145 and 149 were caused by inaccurate registration:

fig 2: subjects 145 and 149 with failed segSubSpace registration. snapshots found here: /autofs/cluster/fsm/users/samseg/tests/2017-03-17_Buckner40/analysis/snapshots

Subject 144 failed samseg with the following error: "maximalDeformation is too small; stopping". The log is located here: /autofs/cluster/fsm/users/samseg/tests/2017-03-17_Buckner40/144/samseg.Y2017.M03.D18.H02.M59.log

With the failed outliers (144, 145, 149) removed, the mean dice coefficients for each label appear better than those in the previous buckner40 test (SamsegEvaluationMarch2016)

fig 3: mean dice scores for each label over all non-outlier Buckner40 subjects

data location

main test directory: /autofs/cluster/fsm/users/samseg/tests/2017-03-17_Buckner40

charts and snapshots: /autofs/cluster/fsm/users/samseg/tests/2017-03-17_Buckner40/analysis

Buckner40 test set data: /autofs/cluster/fsm/users/samseg/subjects/Buckner40

SamsegAsegTesting/2017-03-17_Buckner40 (last edited 2017-03-20 14:01:20 by AndrewHoopes)