Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang yung-jui.yang@yale.edu To: freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" ridatta@mail.med.upenn.edu wrote:
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang yung-jui.yang@yale.edu To: freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
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.
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
We are investigating it. I did fix one thing that you can try if you want - give us your hardware/software info and we will send you a new version of mris_make_surfaces Bruce
P.s. you can also upload and we will take a look
On Apr 10, 2013, at 6:11 AM, "Yang, Daniel" yung-jui.yang@yale.edu wrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" yung-jui.yang@yale.edu wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/18/13 6:27 PM, "Matt Glasser" matt@ma-tea.com wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" ridatta@mail.med.upenn.edu wrote:
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang yung-jui.yang@yale.edu To: freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
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
I have processed more that 600 brains with both versions in the last weeks and the only difference I'm seeing between version 5.2.0 and 5.1, besides the obvious new features, is processing time.
Version 5.2 is 10% faster than 5.1 in an Amazon EC2 instance.
Besides that there's no visible difference in terms of cortical thickness, volumes, etc.
If you have access to computer resources to spare you can run recon-all of both versions in some well known database of images and do a more formal test.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 8:11 AM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" yung-jui.yang@yale.edu wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/18/13 6:27 PM, "Matt Glasser" matt@ma-tea.com wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" ridatta@mail.med.upenn.edu
wrote:
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang yung-jui.yang@yale.edu To: freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
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
Hi PPJ That's exactly what we are doing. Good to hear its stable for you Bruce
On Apr 10, 2013, at 8:38 AM, Pedro Paulo de Magalhães Oliveira Juniorppj@netfilter.com.br wrote:
I have processed more that 600 brains with both versions in the last weeks and the only difference I'm seeing between version 5.2.0 and 5.1, besides the obvious new features, is processing time.
Version 5.2 is 10% faster than 5.1 in an Amazon EC2 instance.
Besides that there's no visible difference in terms of cortical thickness, volumes, etc.
If you have access to computer resources to spare you can run recon-all of both versions in some well known database of images and do a more formal test.
Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 8:11 AM, Yang, Daniel yung-jui.yang@yale.edu wrote: Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" yung-jui.yang@yale.edu wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/18/13 6:27 PM, "Matt Glasser" matt@ma-tea.com wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" ridatta@mail.med.upenn.edu wrote:
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang yung-jui.yang@yale.edu To: freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
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
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
You'll find attached some preliminary data of the comparison we did among versions.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 10:42 AM, Bruce Fischl fischl@nmr.mgh.harvard.eduwrote:
Hi PPJ That's exactly what we are doing. Good to hear its stable for you Bruce
On Apr 10, 2013, at 8:38 AM, Pedro Paulo de Magalhães Oliveira Junior< ppj@netfilter.com.br> wrote:
I have processed more that 600 brains with both versions in the last weeks and the only difference I'm seeing between version 5.2.0 and 5.1, besides the obvious new features, is processing time.
Version 5.2 is 10% faster than 5.1 in an Amazon EC2 instance.
Besides that there's no visible difference in terms of cortical thickness, volumes, etc.
If you have access to computer resources to spare you can run recon-all of both versions in some well known database of images and do a more formal test.
Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 8:11 AM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" yung-jui.yang@yale.edu wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/18/13 6:27 PM, "Matt Glasser" matt@ma-tea.com wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" ridatta@mail.med.upenn.edu
wrote:
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang yung-jui.yang@yale.edu To: freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness
in
particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm
in
5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
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
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi PPJ,
Thanks! It looks interesting. I also found FS 5.2 is faster. Is there any chance you could also provide the cortical thickness of the 2009 atlas (e.g., rh)?
I will take a look into the aseg.volume in my data too.
Best, Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
From: Pedro Paulo de Magalhães Oliveira Junior <ppj@netfilter.com.brmailto:ppj@netfilter.com.br> Date: Wednesday, April 10, 2013 10:49 AM To: Bruce Fischl <fischl@nmr.mgh.harvard.edumailto:fischl@nmr.mgh.harvard.edu> Cc: Daniel Yang <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu>, "freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu" <freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu> Subject: Re: [Freesurfer] Very different results between 5.1.0 and 5.2.0
You'll find attached some preliminary data of the comparison we did among versions.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.brhttp://www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 10:42 AM, Bruce Fischl <fischl@nmr.mgh.harvard.edumailto:fischl@nmr.mgh.harvard.edu> wrote: Hi PPJ That's exactly what we are doing. Good to hear its stable for you Bruce
On Apr 10, 2013, at 8:38 AM, Pedro Paulo de Magalhães Oliveira Junior<ppj@netfilter.com.brmailto:ppj@netfilter.com.br> wrote:
I have processed more that 600 brains with both versions in the last weeks and the only difference I'm seeing between version 5.2.0 and 5.1, besides the obvious new features, is processing time.
Version 5.2 is 10% faster than 5.1 in an Amazon EC2 instance.
Besides that there's no visible difference in terms of cortical thickness, volumes, etc.
If you have access to computer resources to spare you can run recon-all of both versions in some well known database of images and do a more formal test.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.brhttp://www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 8:11 AM, Yang, Daniel <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> wrote: Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454tel:%28203%29%20737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454tel:%28203%29%20737-5454
On 3/18/13 6:27 PM, "Matt Glasser" <matt@ma-tea.commailto:matt@ma-tea.com> wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" <ridatta@mail.med.upenn.edumailto:ridatta@mail.med.upenn.edu> wrote:
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edumailto:ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> To: freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454tel:%28203%29%20737-5454
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto: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.edumailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Ok, I'll try to put together a stat from aparc too.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 12:04 PM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Hi PPJ,
Thanks! It looks interesting. I also found FS 5.2 is faster. Is there any chance you could also provide the cortical thickness of the 2009 atlas (e.g., rh)?
I will take a look into the aseg.volume in my data too.
Best, Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
From: Pedro Paulo de Magalhães Oliveira Junior ppj@netfilter.com.br Date: Wednesday, April 10, 2013 10:49 AM To: Bruce Fischl fischl@nmr.mgh.harvard.edu Cc: Daniel Yang yung-jui.yang@yale.edu, "freesurfer@nmr.mgh.harvard.edu" freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Very different results between 5.1.0 and 5.2.0
You'll find attached some preliminary data of the comparison we did among versions.
Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 10:42 AM, Bruce Fischl <fischl@nmr.mgh.harvard.edu
wrote:
Hi PPJ That's exactly what we are doing. Good to hear its stable for you Bruce
On Apr 10, 2013, at 8:38 AM, Pedro Paulo de Magalhães Oliveira Junior< ppj@netfilter.com.br> wrote:
I have processed more that 600 brains with both versions in the last weeks and the only difference I'm seeing between version 5.2.0 and 5.1, besides the obvious new features, is processing time.
Version 5.2 is 10% faster than 5.1 in an Amazon EC2 instance.
Besides that there's no visible difference in terms of cortical thickness, volumes, etc.
If you have access to computer resources to spare you can run recon-all of both versions in some well known database of images and do a more formal test.
Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 8:11 AM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" yung-jui.yang@yale.edu wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/18/13 6:27 PM, "Matt Glasser" matt@ma-tea.com wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" ridatta@mail.med.upenn.edu
wrote:
I concur. I have seen similar results in primary visual cortex from
~40
subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang yung-jui.yang@yale.edu To: freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness
in
particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm
in
5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
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
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Hi PPJ and all,
I found that the 5.2 – 5.1 difference is primarily seen in the cortical thickness, and much less so in the aseg.volume.
Here, I picked right-amygdala volume as an example of aseg.volume and rh_bankssts_thickness as an example of rh.aparc.thickness.
While the correlation between the two versions of right-amygdala is r = .92 (n = 161), that of the rh_bankssts_thickness is r = .45.
Presumably I believe the correlation should be > .90 for a strong continuity between the two versions?
Do you have anything in the cortical thickness?
Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
From: Pedro Paulo de Magalhães Oliveira Junior <ppj@netfilter.com.brmailto:ppj@netfilter.com.br> Date: Wednesday, April 10, 2013 11:07 AM To: Daniel Yang <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> Cc: Bruce Fischl <fischl@nmr.mgh.harvard.edumailto:fischl@nmr.mgh.harvard.edu>, "freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu" <freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu> Subject: Re: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Ok, I'll try to put together a stat from aparc too.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.brhttp://www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 12:04 PM, Yang, Daniel <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> wrote: Hi PPJ,
Thanks! It looks interesting. I also found FS 5.2 is faster. Is there any chance you could also provide the cortical thickness of the 2009 atlas (e.g., rh)?
I will take a look into the aseg.volume in my data too.
Best, Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454tel:%28203%29%20737-5454
From: Pedro Paulo de Magalhães Oliveira Junior <ppj@netfilter.com.brmailto:ppj@netfilter.com.br> Date: Wednesday, April 10, 2013 10:49 AM To: Bruce Fischl <fischl@nmr.mgh.harvard.edumailto:fischl@nmr.mgh.harvard.edu> Cc: Daniel Yang <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu>, "freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu" <freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu> Subject: Re: [Freesurfer] Very different results between 5.1.0 and 5.2.0
You'll find attached some preliminary data of the comparison we did among versions.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.brhttp://www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 10:42 AM, Bruce Fischl <fischl@nmr.mgh.harvard.edumailto:fischl@nmr.mgh.harvard.edu> wrote: Hi PPJ That's exactly what we are doing. Good to hear its stable for you Bruce
On Apr 10, 2013, at 8:38 AM, Pedro Paulo de Magalhães Oliveira Junior<ppj@netfilter.com.brmailto:ppj@netfilter.com.br> wrote:
I have processed more that 600 brains with both versions in the last weeks and the only difference I'm seeing between version 5.2.0 and 5.1, besides the obvious new features, is processing time.
Version 5.2 is 10% faster than 5.1 in an Amazon EC2 instance.
Besides that there's no visible difference in terms of cortical thickness, volumes, etc.
If you have access to computer resources to spare you can run recon-all of both versions in some well known database of images and do a more formal test.
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.brhttp://www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 8:11 AM, Yang, Daniel <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> wrote: Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454tel:%28203%29%20737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454tel:%28203%29%20737-5454
On 3/18/13 6:27 PM, "Matt Glasser" <matt@ma-tea.commailto:matt@ma-tea.com> wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that highlight the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" <ridatta@mail.med.upenn.edumailto:ridatta@mail.med.upenn.edu> wrote:
I concur. I have seen similar results in primary visual cortex from ~40 subjects. While fs 5.1 estimated mean thickness in the range of 1.5 to 1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3.
Ritobrato Datta, Ph.D. Post Doctoral Researcher Department of Neurology University of Pennsylvania School of Medicine 3rd Floor, Room 312 3710 Hamilton Walk (Goddard Laboratories) Philadelphia, PA 19104-6241 email - ridatta@mail.med.upenn.edumailto:ridatta@mail.med.upenn.edu
----- Original Message ----- From: Daniel Yang <yung-jui.yang@yale.edumailto:yung-jui.yang@yale.edu> To: freesurfer@nmr.mgh.harvard.edumailto:freesurfer@nmr.mgh.harvard.edu Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Dear FreeSurfer Experts,
I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 subjects, and I'm interested in rh_superior_temporal_sulcus_thickness in particular.
Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28 mm in 5.2.0. They are significantly different, t(160) = 56.71.
The correlation between the two versions is r = .33
Is this something possible?? I can't see what went wrong in my procedures.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454tel:%28203%29%20737-5454
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto: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.edumailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
_______________________________________________ Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edumailto:Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Daniel,
Unfortunately, since I run all the individuals again with FreeSurfer 5.2, I just saved from the 5.1 the info I was using for comparison between version - the aseg stats.
I don't have cortical thickness information for 5.1 unless we run all the cases again in 5.1
--------------------------------------------------------------------- Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Thu, Apr 11, 2013 at 2:47 PM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Hi PPJ and all,
I found that the 5.2 – 5.1 difference is primarily seen in the cortical thickness, and much less so in the aseg.volume.
Here, I picked right-amygdala volume as an example of aseg.volume and rh_bankssts_thickness as an example of rh.aparc.thickness.
While the correlation between the two versions of right-amygdala is r = .92 (n = 161), that of the rh_bankssts_thickness is r = .45.
Presumably I believe the correlation should be > .90 for a strong continuity between the two versions?
Do you have anything in the cortical thickness?
Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
From: Pedro Paulo de Magalhães Oliveira Junior ppj@netfilter.com.br Date: Wednesday, April 10, 2013 11:07 AM To: Daniel Yang yung-jui.yang@yale.edu Cc: Bruce Fischl fischl@nmr.mgh.harvard.edu, " freesurfer@nmr.mgh.harvard.edu" freesurfer@nmr.mgh.harvard.edu
Subject: Re: [Freesurfer] Very different results between 5.1.0 and 5.2.0
Ok, I'll try to put together a stat from aparc too.
Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 12:04 PM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Hi PPJ,
Thanks! It looks interesting. I also found FS 5.2 is faster. Is there any chance you could also provide the cortical thickness of the 2009 atlas (e.g., rh)?
I will take a look into the aseg.volume in my data too.
Best, Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
From: Pedro Paulo de Magalhães Oliveira Junior ppj@netfilter.com.br Date: Wednesday, April 10, 2013 10:49 AM To: Bruce Fischl fischl@nmr.mgh.harvard.edu Cc: Daniel Yang yung-jui.yang@yale.edu, "freesurfer@nmr.mgh.harvard.edu" freesurfer@nmr.mgh.harvard.edu Subject: Re: [Freesurfer] Very different results between 5.1.0 and 5.2.0
You'll find attached some preliminary data of the comparison we did among versions.
Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 10:42 AM, Bruce Fischl < fischl@nmr.mgh.harvard.edu> wrote:
Hi PPJ That's exactly what we are doing. Good to hear its stable for you Bruce
On Apr 10, 2013, at 8:38 AM, Pedro Paulo de Magalhães Oliveira Junior< ppj@netfilter.com.br> wrote:
I have processed more that 600 brains with both versions in the last weeks and the only difference I'm seeing between version 5.2.0 and 5.1, besides the obvious new features, is processing time.
Version 5.2 is 10% faster than 5.1 in an Amazon EC2 instance.
Besides that there's no visible difference in terms of cortical thickness, volumes, etc.
If you have access to computer resources to spare you can run recon-all of both versions in some well known database of images and do a more formal test.
Pedro Paulo de Magalhães Oliveira Junior Netfilter & SpeedComm Telecom -- www.netfilter.com.br -- For mobile: http://itunes.apple.com/br/artist/netfilter/id365306441
On Wed, Apr 10, 2013 at 8:11 AM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/19/13 7:07 AM, "Yang, Daniel" yung-jui.yang@yale.edu wrote:
Posting one of the brains.
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
It seems to me that neither version is perfect; however, 5.2.0 is capturing more "black spaces" in the region I'm looking at.
It's in the right hemisphere, TAL coordinate about ~ (44, -46, 20).
Given that the correlation between 5.1.0 and 5.2.0 is r = .33 in the region I examined with my samples, it's not a systematic "predictable" bias.
Any solution?
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 3/18/13 6:27 PM, "Matt Glasser" matt@ma-tea.com wrote:
Do the surfaces look correct in these regions? You might post some screenshots of subjects who have a big difference between 5.1 and 5.2 with the 5.1 and 5.2 white and pial surfaces on volume slices that
highlight
the difference. Without this kind of info, its hard to know which was more correct, 5.1 or 5.2.
Peace,
Matt.
On 3/18/13 5:13 PM, "Ritobrato Datta" ridatta@mail.med.upenn.edu
wrote:
>I concur. I have seen similar results in primary visual cortex from
~40
>subjects. While fs 5.1 estimated mean thickness in the range of 1.5
to
>1.9 in V1, fs 5.2 is giving me V1 thickness in the range of 2 to 2.3. > >Ritobrato Datta, Ph.D. >Post Doctoral Researcher >Department of Neurology >University of Pennsylvania School of Medicine >3rd Floor, Room 312 >3710 Hamilton Walk (Goddard Laboratories) >Philadelphia, PA 19104-6241 >email - ridatta@mail.med.upenn.edu > > >----- Original Message ----- >From: Daniel Yang yung-jui.yang@yale.edu >To: freesurfer@nmr.mgh.harvard.edu >Sent: Mon, 18 Mar 2013 17:44:44 -0400 (EDT) >Subject: [Freesurfer] Very different results between 5.1.0 and 5.2.0 > >Dear FreeSurfer Experts, > >I ran FreeSurfer 5.1.0 and FreeSurfer 5.2.0 on identical set of 161 >subjects, and I'm interested in
rh_superior_temporal_sulcus_thickness in
>particular. > >Previously, the mean thickness is 2.24 mm in 5.1.0; now it is 3.28
mm in
>5.2.0. They are significantly different, t(160) = 56.71. > >The correlation between the two versions is r = .33 > >Is this something possible?? I can't see what went wrong in my >procedures. > >Thanks! >Daniel > >-- >Yung-Jui "Daniel" Yang, PhD >Postdoctoral Researcher >Yale Child Study Center >New Haven, CT >(203) 737-5454 > >_______________________________________________ >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
Freesurfer mailing list Freesurfer@nmr.mgh.harvard.edu https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
Daniel,
We're repeating our paired-analysis of thickness measures between 5.1 and 5.2. In the meantime, to check for correctness, open the brain.finalsurfs.mgz file with the surfaces overlayed, and check the intensity value of the voxels which appear to be non-cortical 'black spaces', relative to neighboring gm voxels. ignore the aseg.mgz gm voxels, as those are not accurate (ie, dont load aseg.mgz when inspecting surfaces, or at least turn if off when inspecting gm regions....its still handy to see where hippocampus sits).
Nick
On Wed, 2013-04-10 at 11:11 +0000, Yang, Daniel wrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
Thanks Nick! I have uploaded the relevant files to you.
Thanks, Daniel
Looking at the image posted previously.....
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
I think it is pretty clear the 5.1 picture has better gray/white contrast. It is a very subtle difference, but you can see it if you look at some pieces of wm that were "missed" by 5.2 in this image.
On Thu, Apr 11, 2013 at 12:33 PM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Thanks Nick! I have uploaded the relevant files to you.
Thanks, Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 4/10/13 1:19 PM, "Nick Schmansky" nicks@nmr.mgh.harvard.edu wrote:
Daniel,
We're repeating our paired-analysis of thickness measures between 5.1 and 5.2. In the meantime, to check for correctness, open the brain.finalsurfs.mgz file with the surfaces overlayed, and check the intensity value of the voxels which appear to be non-cortical 'black spaces', relative to neighboring gm voxels. ignore the aseg.mgz gm voxels, as those are not accurate (ie, dont load aseg.mgz when inspecting surfaces, or at least turn if off when inspecting gm regions....its still handy to see where hippocampus sits).
Nick
On Wed, 2013-04-10 at 11:11 +0000, Yang, Daniel wrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous post below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
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 5.2 image has been smoothed, by a small degree, relative to 5.1. Either prior to FS processing or by FS, it would seem.
On Mon, Apr 15, 2013 at 3:29 AM, Christopher Bell < christopherbell2011@gmail.com> wrote:
Looking at the image posted previously.....
https://yalesurvey.qualtrics.com/SE/?SID=SV_ddwW7I9yMQuCtPn
I think it is pretty clear the 5.1 picture has better gray/white contrast. It is a very subtle difference, but you can see it if you look at some pieces of wm that were "missed" by 5.2 in this image.
On Thu, Apr 11, 2013 at 12:33 PM, Yang, Daniel yung-jui.yang@yale.eduwrote:
Thanks Nick! I have uploaded the relevant files to you.
Thanks, Daniel
-- Yung-Jui "Daniel" Yang, PhD Postdoctoral Researcher Yale Child Study Center New Haven, CT (203) 737-5454
On 4/10/13 1:19 PM, "Nick Schmansky" nicks@nmr.mgh.harvard.edu wrote:
Daniel,
We're repeating our paired-analysis of thickness measures between 5.1 and 5.2. In the meantime, to check for correctness, open the brain.finalsurfs.mgz file with the surfaces overlayed, and check the intensity value of the voxels which appear to be non-cortical 'black spaces', relative to neighboring gm voxels. ignore the aseg.mgz gm voxels, as those are not accurate (ie, dont load aseg.mgz when inspecting surfaces, or at least turn if off when inspecting gm regions....its still handy to see where hippocampus sits).
Nick
On Wed, 2013-04-10 at 11:11 +0000, Yang, Daniel wrote:
Dear FreeSurfer Experts and Users,
Did anyone find similar things using FS 5.2 (please see my previous
post
below)? That is, FS 5.2 is including more non-cortical "black spaces" within pial surfaces, compared to FS 5.1?
I'm not interested in nitpicking but I feel this is a rather serious issue, so I would like to raise it again before it's completely forgotten.
At the meantime I keep receiving Emails from people asking me this issue.
Thanks! Daniel
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
freesurfer@nmr.mgh.harvard.edu
-
Bruce Fischl -
Christopher Bell -
Matt Glasser -
Nick Schmansky -
Pedro Paulo de Magalhães Oliveira Junior -
Ritobrato Datta -
Yang, Daniel