Buckner40 Testing

see also: TestPlan, and Buckner40Notes

Introduction

The 'Buckner40' is a name given to a small subset of subjects originating from Washington University, collected by Randy Buckner and colleagues. It consists of a group of 40 nondemented subjects, ranging in age from 18 to 93 (with no subjects between 30 to 65 years of age), 21 women and 19 men. These data are a subset of a larger open-access structural data set that can be obtained from the fMRIDC at: http://www.fmridc.org/f/fmridc, accession number 2-2004-1168X.

For Freesurfer, this dataset is especially nice as a test-set because of the wide subject age range, and the quality of the scans, which, while good, are not great, but are typical of what Freesurfer has to handle.

The Buckner40Notes wiki page documents the work of examining these brains, slice by slice, to verify their correctness, in terms of how well Freesurfer has reconstructed the white and pial surfaces.

It is important that Freesurfer's recon-all processing stream continue to process all of the subjects successfully, where that means the results are as good as or better than prior 'known-good' results (the reference set). (Note, if Freesurfer improves the results, then a new reference set is created, aka 'reset the standard').

The tests

location of reference set(s)

run on seychelles

use stable and dev

test_recon-all script

what to do if failures are detected