Re: [caret-users] Hi res recon
Dear Donna, Colin, and David, Thank you all for the input. Turning off automatic error correction helped in significantly cutting off processing time. I managed to reconstruct with additional manual labour. It worked reasonably well. The topological errors were not that many, especially when we're only interested in a certain roi. Thanks again everyone. Aditya David Van Essen wrote: >Alternatively, you may find that FreeSurfer will work better for your current >needs. Matt Glasser and others have gotten it to work reasonably well on >macaque structural images. > >David > >On Oct 15, 2014, at 2:38 AM, Donna Dierker wrote: > >> Hi Aditya, >> >> On monkeys, yes. Humans, no. The SureFit algorithm that is in Caret's >> segmentation feature was designed for cubic 1mm human data. It worked >> reasonably well on higher res monkey data, but some of the subroutines will >> likely break on higher res human data (e.g., disconnecting eye, skull, >> hindbrain). >> >> I'd turn all error correction features off and sanity check the initial >> segmentation. If the skull, eye, or hindbrain is still connected, then >> resolving that issue should precede the error correction steps. >> Unfortunately, that will likely take some work. >> >> Donna >> >> >> On Oct 14, 2014, at 5:25 AM, "Dr. Aditya Tri Hernowo, Ph.D" >> wrote: >> >>> Dear users & experts, >>> >>> Does anyone have any experience with reconstructing the cortex on 0.5mm >>> resolution T1 images? I am still having problems with the very long time it >>> takes to perform automatic error correction (more than 3 hours before the >>> software finally crashed). >>> >>> Regards, >>> >>> Aditya Hernowo >>> >>> ___ >>> caret-users mailing list >>> caret-users@brainvis.wustl.edu >>> http://brainvis.wustl.edu/mailman/listinfo/caret-users >> >> >> ___ >> caret-users mailing list >> caret-users@brainvis.wustl.edu >> http://brainvis.wustl.edu/mailman/listinfo/caret-users > > >___ >caret-users mailing list >caret-users@brainvis.wustl.edu >http://brainvis.wustl.edu/mailman/listinfo/caret-users ___ caret-users mailing list caret-users@brainvis.wustl.edu http://brainvis.wustl.edu/mailman/listinfo/caret-users
Re: [caret-users] Hi res recon
Alternatively, you may find that FreeSurfer will work better for your current needs. Matt Glasser and others have gotten it to work reasonably well on macaque structural images. David On Oct 15, 2014, at 2:38 AM, Donna Dierker wrote: > Hi Aditya, > > On monkeys, yes. Humans, no. The SureFit algorithm that is in Caret's > segmentation feature was designed for cubic 1mm human data. It worked > reasonably well on higher res monkey data, but some of the subroutines will > likely break on higher res human data (e.g., disconnecting eye, skull, > hindbrain). > > I'd turn all error correction features off and sanity check the initial > segmentation. If the skull, eye, or hindbrain is still connected, then > resolving that issue should precede the error correction steps. > Unfortunately, that will likely take some work. > > Donna > > > On Oct 14, 2014, at 5:25 AM, "Dr. Aditya Tri Hernowo, Ph.D" > wrote: > >> Dear users & experts, >> >> Does anyone have any experience with reconstructing the cortex on 0.5mm >> resolution T1 images? I am still having problems with the very long time it >> takes to perform automatic error correction (more than 3 hours before the >> software finally crashed). >> >> Regards, >> >> Aditya Hernowo >> >> ___ >> caret-users mailing list >> caret-users@brainvis.wustl.edu >> http://brainvis.wustl.edu/mailman/listinfo/caret-users > > > ___ > caret-users mailing list > caret-users@brainvis.wustl.edu > http://brainvis.wustl.edu/mailman/listinfo/caret-users ___ caret-users mailing list caret-users@brainvis.wustl.edu http://brainvis.wustl.edu/mailman/listinfo/caret-users
Re: [caret-users] Hi res recon
Hi Aditya, On monkeys, yes. Humans, no. The SureFit algorithm that is in Caret's segmentation feature was designed for cubic 1mm human data. It worked reasonably well on higher res monkey data, but some of the subroutines will likely break on higher res human data (e.g., disconnecting eye, skull, hindbrain). I'd turn all error correction features off and sanity check the initial segmentation. If the skull, eye, or hindbrain is still connected, then resolving that issue should precede the error correction steps. Unfortunately, that will likely take some work. Donna On Oct 14, 2014, at 5:25 AM, "Dr. Aditya Tri Hernowo, Ph.D" wrote: > Dear users & experts, > > Does anyone have any experience with reconstructing the cortex on 0.5mm > resolution T1 images? I am still having problems with the very long time it > takes to perform automatic error correction (more than 3 hours before the > software finally crashed). > > Regards, > > Aditya Hernowo > > ___ > caret-users mailing list > caret-users@brainvis.wustl.edu > http://brainvis.wustl.edu/mailman/listinfo/caret-users ___ caret-users mailing list caret-users@brainvis.wustl.edu http://brainvis.wustl.edu/mailman/listinfo/caret-users
