Re: [Freesurfer] parellizing Freesurfer
Thank you, Pedro. That is really useful. 2011/2/27 Pedro Paulo de Magalhães Oliveira Junior > You'll find some answers here: > http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg15728.html > > > > 2011/2/26 Pablo Polosecki > > Hi Pedro and everyone, >> >> Thanks so much for pointing me there. I seems like the only way to >> parallelize is to run several recon-all simultanously intstead of wating for >> one to finshe and then run the following. I wonder if you (of someone else) >> managed to achieve some further decrease in running times, making a single >> recon-all run faster or something like that. Was it any different in the >> Amazon EC2? >> >> It would seem that the analysis fo functional data (at least parts of it) >> could run in parallel, since the analysis of one voxel is independent of the >> rest. I would love to know if this is something people do. An ugly >> implementation could be to split the volumes into several small volume files >> and run the analysis separately and then merge back the output volumes. Has >> anyone done something like that or nicer? >> >> Thanks! >> >> Pablo >> >> 2011/2/26 Pedro Paulo de Magalhães Oliveira Junior >> >> Yes. Take a look at: >>> http://surfer.nmr.mgh.harvard.edu/fswiki/UserContributions/FAQ#Q.HowcanIreducethetimeofrecon-allinagroupofpatients.3F >>> >>> Also I have successfully executed it at Amazon EC2 instances. >>> >>> PPJ >>> - >>> Pedro Paulo de Magalhães Oliveira Junior >>> Diretor de Operações >>> Netfilter & SpeedComm Telecom >>> -- www.netfilter.com.br >>> -- For mobile: http://www.netfilter.com.br/mobile >>> >>> >>> >>> >>> On Sat, Feb 26, 2011 at 19:59, Pablo Polosecki < >>> ppolose...@mail.rockefeller.edu> wrote: >>> Hi all, I was wondering if someone out there has any experience trying to parallelize their analysis streams in clusters or multiprocessor machines. I was thinking also about the cloud computing services offered by Google, Amazon or other. Has anyone tried or has any plans of implementing something along those lines? Best wishes, Pablo ___ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail. >>> >> > ___ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
Re: [Freesurfer] parellizing Freesurfer
You'll find some answers here: http://www.mail-archive.com/freesurfer@nmr.mgh.harvard.edu/msg15728.html 2011/2/26 Pablo Polosecki > Hi Pedro and everyone, > > Thanks so much for pointing me there. I seems like the only way to > parallelize is to run several recon-all simultanously intstead of wating for > one to finshe and then run the following. I wonder if you (of someone else) > managed to achieve some further decrease in running times, making a single > recon-all run faster or something like that. Was it any different in the > Amazon EC2? > > It would seem that the analysis fo functional data (at least parts of it) > could run in parallel, since the analysis of one voxel is independent of the > rest. I would love to know if this is something people do. An ugly > implementation could be to split the volumes into several small volume files > and run the analysis separately and then merge back the output volumes. Has > anyone done something like that or nicer? > > Thanks! > > Pablo > > 2011/2/26 Pedro Paulo de Magalhães Oliveira Junior > > Yes. Take a look at: >> http://surfer.nmr.mgh.harvard.edu/fswiki/UserContributions/FAQ#Q.HowcanIreducethetimeofrecon-allinagroupofpatients.3F >> >> Also I have successfully executed it at Amazon EC2 instances. >> >> PPJ >> - >> Pedro Paulo de Magalhães Oliveira Junior >> Diretor de Operações >> Netfilter & SpeedComm Telecom >> -- www.netfilter.com.br >> -- For mobile: http://www.netfilter.com.br/mobile >> >> >> >> >> On Sat, Feb 26, 2011 at 19:59, Pablo Polosecki < >> ppolose...@mail.rockefeller.edu> wrote: >> >>> Hi all, >>> >>> I was wondering if someone out there has any experience trying to >>> parallelize their analysis streams in clusters or multiprocessor machines. I >>> was thinking also about the cloud computing services offered by Google, >>> Amazon or other. Has anyone tried or has any plans of implementing something >>> along those lines? >>> >>> Best wishes, >>> >>> Pablo >>> >>> >>> >>> >>> ___ >>> Freesurfer mailing list >>> Freesurfer@nmr.mgh.harvard.edu >>> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >>> >>> >>> The information in this e-mail is intended only for the person to whom it >>> is >>> addressed. If you believe this e-mail was sent to you in error and the >>> e-mail >>> contains patient information, please contact the Partners Compliance >>> HelpLine at >>> http://www.partners.org/complianceline . If the e-mail was sent to you >>> in error >>> but does not contain patient information, please contact the sender and >>> properly >>> dispose of the e-mail. >>> >>> >> > ___ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer The information in this e-mail is intended only for the person to whom it is addressed. If you believe this e-mail was sent to you in error and the e-mail contains patient information, please contact the Partners Compliance HelpLine at http://www.partners.org/complianceline . If the e-mail was sent to you in error but does not contain patient information, please contact the sender and properly dispose of the e-mail.
Re: [Freesurfer] fixing hippocampus segmentation
Dear Bruce, sorry for late answer. I uploaded the subject here http://www.sendspace.com/file/od7axn. I would appreciate if you could write how I can control segmentation if I need to do it. Should I try different settings of mri_ca_label? Thank you very much for your time, Tanja. On Wed, Feb 23, 2011 at 4:59 PM, Bruce Fischl wrote: > hmm, that is large, but the tiny one-voxel border you showed in your image > wouldn't come close to explaining that. Do you want to upload the subject > and we'll take a look? > Bruce > > > On Wed, 23 Feb 2011, Tetiana Dadakova wrote: > >> Dear Bruce, >> >> The problem is that the volumes I get for hippocampus are much larger >> than they should be (like 5000 mm^3 for healthy female). Same is for >> some other structures. >> >> Best wishes, >> Tanja. >> >> >> On Wed, Feb 23, 2011 at 3:08 PM, Bruce Fischl >> wrote: >>> >>> Hi Tetiana, >>> >>> I think what you're seeing are just the consequences of a voxel-based >>> segmentation. The borders have to be voxelated. Note that much of this >>> will >>> be corrected in the volume calculation that performs a partial-volume >>> correction. >>> >>> cheers >>> Bruce >>> >>> On Wed, 23 Feb 2011, Tetiana Dadakova wrote: >>> Dear Bruce, dear all, I attach several images to show where I would like to manually interfere into segmentation process. img1: Hippocampus' volume covers part of amygdala and parahippocampal area. The intensity in these non-hippocampal regions are between 75-95. img2: Segmentation of brain stem (problem point are shown by arrows). This problem repeats more or less for every subject. img3: Edges of thalamus are too "pixelated". I mean, shouldn't the line be smoother? Can this cause overestimation of thalamus' volume? Manual correction (drawing) of all the mis-segmented parts would take a lot of time, so I would like to find out other way of controlling the segmentation process. (Something like "putting control points" would be perfect). Thank you, Tanja. On Mon, Feb 21, 2011 at 2:56 PM, Bruce Fischl wrote: > > Hi Tanja, > Can you send us an image? Is parahippocampal white matter significantly > darker than 110? > Bruce > > > > On Feb 21, 2011, at 6:41 AM, Tetiana Dadakova > wrote: > >> Dear Freesurfer users, >> >> For some of my subjects Freesurfer strongly overestimates hippocampus' >> volume. Is there any way to fix the segmentation of hippocampus by >> e.g. adding some control points (like for white matter here >> http://surfer.nmr.mgh.harvard.edu/fswiki/FsTutorial/ControlPoints)? >> >> Thank you, >> Tanja. >> ___ >> Freesurfer mailing list >> Freesurfer@nmr.mgh.harvard.edu >> https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer >> >> > >>> >>> >>> The information in this e-mail is intended only for the person to whom it >>> is >>> addressed. If you believe this e-mail was sent to you in error and the >>> e-mail >>> contains patient information, please contact the Partners Compliance >>> HelpLine at >>> http://www.partners.org/complianceline . If the e-mail was sent to you in >>> error >>> but does not contain patient information, please contact the sender and >>> properly >>> dispose of the e-mail. >>> >>> >> >> >> > ___ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer