Hi Stephanie
I see. Yes, I guess this is ok, you just want to be careful when tossing
datasets that you aren’t biasing your results. Always better to look at them if
you can. Big errors are pretty obvious and can be found quite rapidly (e.g. by
zipping through a bunch of snapshots with nmovie_qt)
Cheers
Bruce
From: freesurfer-boun...@nmr.mgh.harvard.edu
<freesurfer-boun...@nmr.mgh.harvard.edu> On Behalf Of Stephanie K
Sent: Friday, November 20, 2020 6:25 AM
To: Freesurfer support list <freesurfer@nmr.mgh.harvard.edu>
Subject: Re: [Freesurfer] Mean thickness estimation {Disarmed}
External Email - Use Caution
Hi Bruce,
No, I don’t think that you’ve misunderstood. I was asking about cohorts that
had not been visually inspected in my first message and then in my last message
I wanted to check that I didn’t need to also remove people from visually
inspected data based on statistical outliers (as a further step). In
conclusion, from what I’ve understood: if there is no visual inspection of
data, it is best to remove statistical outliers (based on standard deviations)
even for global measures such estimated intracranial volume, mean thickness
from left and right hemispheres. When data is visually inspected, there is no
need to remove any data based on statistical outliers even if those exist as
these should be good quality. For the visual inspected data, I have kept all
participants (which have an acceptable cortical surface reconstruction rating)
for the regional variables and I also retained all participants for global
measures (thickness of mean hemisphere, estimated intracranial volume) as the
data had been processed and removed from further Freesurfer processing if they
had severe artifacts or irregularities.
Many thanks for your advice.
On Thursday, November 19, 2020, Fischl, Bruce
<bfis...@mgh.harvard.edu<mailto:bfis...@mgh.harvard.edu>> wrote:
Hi Stephanie
Sorry, I think I misunderstood. If you have visually inspected the analysis
results and think that they are accurate then you definitely should leave those
subjects in
Cheers
Bruce
From:
freesurfer-boun...@nmr.mgh.harvard.edu<mailto:freesurfer-boun...@nmr.mgh.harvard.edu>
<freesurfer-boun...@nmr.mgh.harvard.edu<mailto:freesurfer-boun...@nmr.mgh.harvard.edu>>
On Behalf Of Stephanie K
Sent: Thursday, November 19, 2020 4:19 PM
To: Freesurfer support list
<freesurfer@nmr.mgh.harvard.edu<mailto:freesurfer@nmr.mgh.harvard.edu>>
Subject: Re: [Freesurfer] Mean thickness estimation {Disarmed}
External Email - Use Caution
Hi Bruce,
Is it also necessary to remove outliers (eg. 3 SD from the mean) for
Freesurfer structural measures of MRI images that have been visually inspected
as a further data cleaning step?
I really appreciate your help.
Thanks
On Thu, Nov 19, 2020 at 3:55 PM Fischl, Bruce
<bfis...@mgh.harvard.edu<mailto:bfis...@mgh.harvard.edu>> wrote:
Hi Stephanie
I think that is always a good idea, unless for some reason it isn’t possible
Cheers
Bruce
From:
freesurfer-boun...@nmr.mgh.harvard.edu<mailto:freesurfer-boun...@nmr.mgh.harvard.edu>
<freesurfer-boun...@nmr.mgh.harvard.edu<mailto:freesurfer-boun...@nmr.mgh.harvard.edu>>
On Behalf Of Stephanie K
Sent: Thursday, November 19, 2020 2:14 AM
To: freesurfer@nmr.mgh.harvard.edu<mailto:freesurfer@nmr.mgh.harvard.edu>
Subject: Re: [Freesurfer] Mean thickness estimation
External Email - Use Caution
Hi Bruce,
Thanks for the prompt reply. It is my understanding that l_thickness,
r_thickness and estimated intracranial volume are accurately measured. Would I
still need to identify and remove outliers if visual inspection of the images
has not been done?
Many thanks,
Stephanie
On Wed, Nov 18, 2020 at 5:24 PM Stephanie K
<rklin...@gmail.com<mailto:rklin...@gmail.com>> wrote:
Hi,
I want to estimate the mean cortical thickness. For this I have summed the
thickness across all 34 regions mapped to the Desikan-Killiany atlas. However,
I also have the average mean thickness of left and right hemispheres (direct
output variables of Freesurfer). As there is no visual inspection of the
imaging in the particular cohort, I remove measures that are 3 standard
deviations above or below the mean. Hence, I may expect more outliers to be
removed when I take the average across the regions. I am using these brain
measures as outcomes in association analyses with the genetic score as the
exposure. For the mean thickness (averaged across the left and right hemisphere
thickness variables of freesurfer after removing outliers), the regression
coefficients have a smaller standard deviation than with thickness averaged
across the 34 regions. I’m not sure which one to use - which one is more
accurate? When I look at the mean thickness (which I derived using 34 regions)
and it’s standard deviation, it is similar to that of the average mean
thickness across the two hemispheres as well as the standard deviation of that.
Can you suggest what is most accurate please and what the difference is between
the mean thickness across the two hemispheres obtained from freesurfer and
those calculated across the regions? Why does one result in more precision than
in the other?
Thank you!
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu<mailto:Freesurfer@nmr.mgh.harvard.edu>
MailScanner has detected a possible fraud attempt from "secure-web.cisco.com"
claiming to be
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer<https://secure-web.cisco.com/1KPUR-tLJEURgzYmremNK46BwlQzhr2HuE-Vgj0d_IfX_gecAUGN6QjCWYLIn3brTZNmVndmFck4JXbECP-4ChBCCB9on11GCIcDXo_CeNsBG78TW3Un3dAkQZk9obRuWPnrYHXzp4kS7t-Y4GsaRBH04naRD0rppbxKez7XlIyHvl1NMwlQaEMRHOk6FqYoTIVHV6UrlC5oR6DZFRClnWLATwzWDCvj5zCYsdJ7pt7xmr1fz3Jml3iS6BjOivmfV4vWBZ-x-rpFZ_zoHRv07WA/https%3A%2F%2Fmail.nmr.mgh.harvard.edu%2Fmailman%2Flistinfo%2Ffreesurfer>
_______________________________________________
Freesurfer mailing list
Freesurfer@nmr.mgh.harvard.edu
https://mail.nmr.mgh.harvard.edu/mailman/listinfo/freesurfer
