Hi freesurfer members,
I have a T1 (with contrast) sprg 1mm sequencing MRI for a few patients. Recon-all runs to completion, with no errors, but the pial surfaces look quite bad (see pial.jpghttps://drive.google.com/open?id=172UOGsrYDnMInQNEDcj6LQ8IUxXNMFiR). Some notes: The tessellation steps take a long time – usually there are ~200 defects to correct, and 1 or 2 defects have 20000+ vertices. Because of the contrast agent, there is a layer of dura and blood vessels visible after the skull strip, perhaps this is causing issues with gray matter segmentation (see slice_with_pial.pnghttps://drive.google.com/open?id=1FExPxgPN4x-bvxp3wAYx9mR9RXQ7vJ1M). The visible contrast between gray and white matter is very low for these MRIs.
What I’ve tried so far, but did not help Define white matter control points as seeds into the auto-recon2 phase. Adjust (lower) watershed_preflood_heights for the skull stripping process. The dura still remain, and it appears that some of the throat is still present, but I’m not sure that is the main issue.
Because the tessellation step takes 30+ hours to run, it’s too difficult to do trial-and-error parameter adjustments and see their effects on the outcome. So I am hoping to get some advice and direction. Are my MRI scans too poor quality for freesurfer to reconstruct a 3D model from?
link to raw data https://gate.nmr.mgh.harvard.edu/filedrop2/index.php?p=3xqqbhtsgsp https://drive.google.com/open?id=1PWzrKZ37RN08rxy_Nv7mMn1sxTo2nyao
Thanks,
Seth Foster Research Assistant, Cogan Lab Department of Neurosurgery Duke University seth.foster@duke.edumailto:seth.foster@duke.edu