Re: [caret-users] threshold of metric file
OK. Thanks, Donna. Zhiwei At 2014-10-30 22:29:32, "Donna Dierker" wrote: >I'm sorry I don't know the answer to the Freesurfer question, but someone else >might know. And I confess I don't understand the rationale for thresholding >at that value, possibly because I am unfamiliiar with the contents of those >files. > >So I'm keeping the responsibility for ensuring a reasonable threshold on you. >;-) But If it's each subject, then it's probably worth scripting it using >these caret_command tools: > > caret_command -surface-region-of-interest-selection > [-metric metric-file-name column min max SEL-TYPE] > > caret_command -surface-roi-statistical-report > >And you can use each subject's surface/topo for that surface area calculation. > Your min is your threshold and max something like 999. Your report will >include the area of the suprathreshold regions, and you can grep that line >from the resulting report file. > > >On Oct 29, 2014, at 11:30 PM, wangzhiwei3233 wrote: > >> Hi, Donna, >> >> My purpose is counting activated area size on each subject. >> >> I did do significance test using Freesurfer on individual level. The results >> contained many files,for example, "sig.nii.gz" and "Fsig.nii.gz" >> corresponding to result of t and f test respectively. Is that right? >> But I do not know how to determine tha suprathreshold on subject level as >> you mentioned. >> >> For display and counting area size, I converted the results file(sig.nii.gz) >> to Caret. Then I count area size on Caret. When I counted area size, I >> selected a uniform threshold 1.3, i.e.-log10(0.05) for each subject. So I >> set the scale to 1.3 ~ maxminum on Caret. Is this right? >> >> So I could draw a border around the suprathreshold region and generated a >> paint file. I got the area size of the region using the paint file. >> >> Is there any step wrong? >> >> Thanks! >> Zhiwei >> >> >> >> At 2014-10-30 00:27:14, "Donna Dierker" wrote: >> >Hi wangzhiwei, >> > >> >I'm a little confused by the question. There mention of area size and >> >scales hints that there might be a confusion between tools used for >> >quantification / significance testing and those used for display purposes. >> > >> >Freesurfer has its own tools for significance testing, so you could use >> >those. We often use threshold-free cluster enhancement for that purpose, >> >which finds the significance threshold. Suprathreshold area can be >> >computed once this threshold has been determined. >> > >> >But usually when I make a figure, I generate border around the >> >suprathreshold regions and display these bordersover the "real" t- or >> >f-map, using a scale that corresponds to my alpha (e.g., .05) divided by >> >two (since I usually do both right and left hem tests). I compute this t >> >or f-stat using my n / degrees of freedom. >> > >> >So the significance testing and display steps are separate, the way I do it. >> > >> >Now you might not be going as far as significance testing. Sometimes you >> >just want to look at some preliminary data -- particularly for a single >> >subject. A good start might be to understand if this is a single subject, >> >group results, what kind of statistic. >> > >> >And certainly not everyone does this the way I do, so it would be helpful >> >for others to weigh in with their viewpoints/conventions. >> > >> >Donna >> > >> > >> >On Oct 28, 2014, at 9:53 PM, wangzhiwei3233 wrote: >> > >> >> Hi, experts, >> >> I converted fMRI results derived from freesurfer to caret, and not I want >> >> to count activation areas on caret. So there is a problem of threshold >> >> and scale. >> >> >> >> Auto scale range is 0~60. I found that the area size was different when >> >> using scale 1.3~4 from when using scale 1.3~60. And the latter one was >> >> smaller. However , in the latter case(1.3~60), the value of a point that >> >> was next to the border of activation area but in non-activation area was >> >> a little bit lager than the threshold 1.3. >> >> >> >> How to set the scale to guarantee the activation accurate? >> >> >> >> Best! >> >> >> >> >> >> ___ >> >> 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 ___ caret-users mailing list caret-users@brainvis.wustl.edu
Re: [caret-users] threshold of metric file
I'm sorry I don't know the answer to the Freesurfer question, but someone else might know. And I confess I don't understand the rationale for thresholding at that value, possibly because I am unfamiliiar with the contents of those files. So I'm keeping the responsibility for ensuring a reasonable threshold on you. ;-) But If it's each subject, then it's probably worth scripting it using these caret_command tools: caret_command -surface-region-of-interest-selection [-metric metric-file-name column min max SEL-TYPE] caret_command -surface-roi-statistical-report And you can use each subject's surface/topo for that surface area calculation. Your min is your threshold and max something like 999. Your report will include the area of the suprathreshold regions, and you can grep that line from the resulting report file. On Oct 29, 2014, at 11:30 PM, wangzhiwei3233 wrote: > Hi, Donna, > > My purpose is counting activated area size on each subject. > > I did do significance test using Freesurfer on individual level. The results > contained many files,for example, "sig.nii.gz" and "Fsig.nii.gz" > corresponding to result of t and f test respectively. Is that right? > But I do not know how to determine tha suprathreshold on subject level as you > mentioned. > > For display and counting area size, I converted the results file(sig.nii.gz) > to Caret. Then I count area size on Caret. When I counted area size, I > selected a uniform threshold 1.3, i.e.-log10(0.05) for each subject. So I set > the scale to 1.3 ~ maxminum on Caret. Is this right? > > So I could draw a border around the suprathreshold region and generated a > paint file. I got the area size of the region using the paint file. > > Is there any step wrong? > > Thanks! > Zhiwei > > > > At 2014-10-30 00:27:14, "Donna Dierker" wrote: > >Hi wangzhiwei, > > > >I'm a little confused by the question. There mention of area size and > >scales hints that there might be a confusion between tools used for > >quantification / significance testing and those used for display purposes. > > > >Freesurfer has its own tools for significance testing, so you could use > >those. We often use threshold-free cluster enhancement for that purpose, > >which finds the significance threshold. Suprathreshold area can be computed > >once this threshold has been determined. > > > >But usually when I make a figure, I generate border around the > >suprathreshold regions and display these bordersover the "real" t- or f-map, > >using a scale that corresponds to my alpha (e.g., .05) divided by two (since > >I usually do both right and left hem tests). I compute this t or f-stat > >using my n / degrees of freedom. > > > >So the significance testing and display steps are separate, the way I do it. > > > >Now you might not be going as far as significance testing. Sometimes you > >just want to look at some preliminary data -- particularly for a single > >subject. A good start might be to understand if this is a single subject, > >group results, what kind of statistic. > > > >And certainly not everyone does this the way I do, so it would be helpful > >for others to weigh in with their viewpoints/conventions. > > > >Donna > > > > > >On Oct 28, 2014, at 9:53 PM, wangzhiwei3233 wrote: > > > >> Hi, experts, > >> I converted fMRI results derived from freesurfer to caret, and not I want > >> to count activation areas on caret. So there is a problem of threshold and > >> scale. > >> > >> Auto scale range is 0~60. I found that the area size was different when > >> using scale 1.3~4 from when using scale 1.3~60. And the latter one was > >> smaller. However , in the latter case(1.3~60), the value of a point that > >> was next to the border of activation area but in non-activation area was a > >> little bit lager than the threshold 1.3. > >> > >> How to set the scale to guarantee the activation accurate? > >> > >> Best! > >> > >> > >> ___ > >> 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] threshold of metric file
Hi, Donna, My purpose is counting activated area size on each subject. I did do significance test using Freesurfer on individual level. The results contained many files,for example, "sig.nii.gz" and "Fsig.nii.gz" corresponding to result of t and f test respectively. Is that right? But I do not know how to determine tha suprathreshold on subject level as you mentioned. For display and counting area size, I converted the results file(sig.nii.gz) to Caret. Then I count area size on Caret. When I counted area size, I selected a uniform threshold 1.3, i.e.-log10(0.05) for each subject. So I set the scale to 1.3 ~ maxminum on Caret. Is this right? So I could draw a border around the suprathreshold region and generated a paint file. I got the area size of the region using the paint file. Is there any step wrong? Thanks! Zhiwei At 2014-10-30 00:27:14, "Donna Dierker" wrote: >Hi wangzhiwei, > >I'm a little confused by the question. There mention of area size and scales >hints that there might be a confusion between tools used for quantification / >significance testing and those used for display purposes. > >Freesurfer has its own tools for significance testing, so you could use those. > We often use threshold-free cluster enhancement for that purpose, which finds >the significance threshold. Suprathreshold area can be computed once this >threshold has been determined. > >But usually when I make a figure, I generate border around the suprathreshold >regions and display these bordersover the "real" t- or f-map, using a scale >that corresponds to my alpha (e.g., .05) divided by two (since I usually do >both right and left hem tests). I compute this t or f-stat using my n / >degrees of freedom. > >So the significance testing and display steps are separate, the way I do it. > >Now you might not be going as far as significance testing. Sometimes you just >want to look at some preliminary data -- particularly for a single subject. A >good start might be to understand if this is a single subject, group results, >what kind of statistic. > >And certainly not everyone does this the way I do, so it would be helpful for >others to weigh in with their viewpoints/conventions. > >Donna > > >On Oct 28, 2014, at 9:53 PM, wangzhiwei3233 wrote: > >> Hi, experts, >> I converted fMRI results derived from freesurfer to caret, and not I want to >> count activation areas on caret. So there is a problem of threshold and >> scale. >> >> Auto scale range is 0~60. I found that the area size was different when >> using scale 1.3~4 from when using scale 1.3~60. And the latter one was >> smaller. However , in the latter case(1.3~60), the value of a point that >> was next to the border of activation area but in non-activation area was a >> little bit lager than the threshold 1.3. >> >> How to set the scale to guarantee the activation accurate? >> >> Best! >> >> >> ___ >> 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] threshold of metric file
Hi wangzhiwei, I'm a little confused by the question. There mention of area size and scales hints that there might be a confusion between tools used for quantification / significance testing and those used for display purposes. Freesurfer has its own tools for significance testing, so you could use those. We often use threshold-free cluster enhancement for that purpose, which finds the significance threshold. Suprathreshold area can be computed once this threshold has been determined. But usually when I make a figure, I generate border around the suprathreshold regions and display these bordersover the "real" t- or f-map, using a scale that corresponds to my alpha (e.g., .05) divided by two (since I usually do both right and left hem tests). I compute this t or f-stat using my n / degrees of freedom. So the significance testing and display steps are separate, the way I do it. Now you might not be going as far as significance testing. Sometimes you just want to look at some preliminary data -- particularly for a single subject. A good start might be to understand if this is a single subject, group results, what kind of statistic. And certainly not everyone does this the way I do, so it would be helpful for others to weigh in with their viewpoints/conventions. Donna On Oct 28, 2014, at 9:53 PM, wangzhiwei3233 wrote: > Hi, experts, > I converted fMRI results derived from freesurfer to caret, and not I want to > count activation areas on caret. So there is a problem of threshold and scale. > > Auto scale range is 0~60. I found that the area size was different when using > scale 1.3~4 from when using scale 1.3~60. And the latter one was smaller. > However , in the latter case(1.3~60), the value of a point that was next to > the border of activation area but in non-activation area was a little bit > lager than the threshold 1.3. > > How to set the scale to guarantee the activation accurate? > > Best! > > > ___ > 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] threshold of metric file
Hi, experts, I converted fMRI results derived from freesurfer to caret, and not I want to count activation areas on caret. So there is a problem of threshold and scale. Auto scale range is 0~60. I found that the area size was different when using scale 1.3~4 from when using scale 1.3~60. And the latter one was smaller. However , in the latter case(1.3~60), the value of a point that was next to the border of activation area but in non-activation area was a little bit lager than the threshold 1.3. How to set the scale to guarantee the activation accurate? Best!___ caret-users mailing list caret-users@brainvis.wustl.edu http://brainvis.wustl.edu/mailman/listinfo/caret-users
