## marsbar-users

 [Marsbar-users] Question about circularity From: Sharon Gilaie-Dotan - 2013-06-28 21:17:18 Attachments: Message as HTML ```Hi, I am using a VBM model, say with predictors A, B, and C. The design matrix I am loading to Marsbar is the one that I have created with the model. Now I am sampling with Marsbar regions that are completely independent from the model (anatomically defined ROIs). I am sampling with Marsbar using "Extract ROI data (default)" option. I thought that I was extracting GM volume from these ROIs basically independent of the model I defined, but is that really the case? I.e. are the sampled values independent from A, B, or C, or are they in some way dependent on them? BTW to check I also defined a model with predictors A, B,C,and D and the sampled values Marsbar gives me are identical to my original model, so I am assuming independence from the model predictors, but would still wish to know for sure. Thanks! -- Sharon Gilaie-Dotan, PhD Institute of Cognitive Neuroscience University College London Alexandra House 17 Queen Square London WC1N 3AR Tel: +44-20-76791122 Mob: +44-78-1450-3536 Mob: +972-54-4736181 ```
 Re: [Marsbar-users] Question about circularity From: Matthew Brett - 2013-07-01 10:22:42 ```Hi, On Fri, Jun 28, 2013 at 10:17 PM, Sharon Gilaie-Dotan wrote: > Hi, > I am using a VBM model, say with predictors A, B, and C. > The design matrix I am loading to Marsbar is the one that I have created > with the model. > Now I am sampling with Marsbar regions that are completely independent from > the model (anatomically defined ROIs). I am sampling with Marsbar using > "Extract ROI data (default)" option. I thought that I was extracting GM > volume from these ROIs basically independent of the model I defined, but is > that really the case? > I.e. are the sampled values independent from A, B, or C, or are they in some > way dependent on them? The sampled values don't depend on the design, they only use the images and the scaling factors in the SPM design. These in turn only depend only on the mean signal in the images, not the design. So, yes, they should be independent. > BTW to check I also defined a model with predictors A, B,C,and D and the > sampled values Marsbar gives me are identical to my original model, so I am > assuming independence from the model predictors, but would still wish to > know for sure. Thanks! Best, Matthew ```
 Re: [Marsbar-users] Question about circularity From: Sharon Gilaie-Dotan - 2013-07-01 14:40:08 Attachments: Message as HTML ```thanks so much for this. Sharon ---- Sharon Gilaie-Dotan, PhD Institute of Cognitive Neuroscience & Wellcome Trust Centre for NeuroimagingUniversity College London Alexandra House 17 Queen Square London WC1N 3AR Tel: +44-20-76791122 Mob: +44-78-1450-3536 Mob: +972-54-4736181 On Jul 1, 2013 12:22 PM, "Matthew Brett" wrote: > Hi, > > On Fri, Jun 28, 2013 at 10:17 PM, Sharon Gilaie-Dotan > wrote: > > Hi, > > I am using a VBM model, say with predictors A, B, and C. > > The design matrix I am loading to Marsbar is the one that I have created > > with the model. > > Now I am sampling with Marsbar regions that are completely independent > from > > the model (anatomically defined ROIs). I am sampling with Marsbar using > > "Extract ROI data (default)" option. I thought that I was extracting GM > > volume from these ROIs basically independent of the model I defined, but > is > > that really the case? > > I.e. are the sampled values independent from A, B, or C, or are they in > some > > way dependent on them? > > The sampled values don't depend on the design, they only use the > images and the scaling factors in the SPM design. These in turn only > depend only on the mean signal in the images, not the design. So, > yes, they should be independent. > > > BTW to check I also defined a model with predictors A, B,C,and D and the > > sampled values Marsbar gives me are identical to my original model, so I > am > > assuming independence from the model predictors, but would still wish to > > know for sure. Thanks! > > Best, > > Matthew > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by Windows: > > Build for Windows Store. > > http://p.sf.net/sfu/windows-dev2dev > _______________________________________________ > Marsbar-users mailing list > Marsbar-users@... > https://lists.sourceforge.net/lists/listinfo/marsbar-users > ```
 Re: [Marsbar-users] Question about circularity From: Sharon Gilaie-Dotan - 2013-11-15 17:19:09 Attachments: Message as HTML ```Dear Matthew, With respect to the question I have asked a while ago, I re-read your answer and want to understand one more bit (if I may) - you wrote that "The sampled values don't depend on the design, they only use the images and the scaling factors in the SPM design. These in turn only *depend only on the mean signal in the images*, not the design. So, yes, they should be independent." Can I then ask what you refer to when you write the mean signal in the images. Does it relate to the mean signal of each participant's image (i.e. the mean GM intensity of that image of that participant) - or perhaps the mean signal in a specific voxel across participants? or perhaps something else? Thanks so much, Sharon Gilaie-Dotan On Mon, Jul 1, 2013 at 1:22 PM, Matthew Brett wrote: > Hi, > > On Fri, Jun 28, 2013 at 10:17 PM, Sharon Gilaie-Dotan > wrote: > > Hi, > > I am using a VBM model, say with predictors A, B, and C. > > The design matrix I am loading to Marsbar is the one that I have created > > with the model. > > Now I am sampling with Marsbar regions that are completely independent > from > > the model (anatomically defined ROIs). I am sampling with Marsbar using > > "Extract ROI data (default)" option. I thought that I was extracting GM > > volume from these ROIs basically independent of the model I defined, but > is > > that really the case? > > I.e. are the sampled values independent from A, B, or C, or are they in > some > > way dependent on them? > > The sampled values don't depend on the design, they only use the > images and the scaling factors in the SPM design. These in turn only > depend only on the mean signal in the images, not the design. So, > yes, they should be independent. > > > BTW to check I also defined a model with predictors A, B,C,and D and the > > sampled values Marsbar gives me are identical to my original model, so I > am > > assuming independence from the model predictors, but would still wish to > > know for sure. Thanks! > > Best, > > Matthew > > > ------------------------------------------------------------------------------ > This SF.net email is sponsored by Windows: > > Build for Windows Store. > > http://p.sf.net/sfu/windows-dev2dev > _______________________________________________ > Marsbar-users mailing list > Marsbar-users@... > https://lists.sourceforge.net/lists/listinfo/marsbar-users > -- Sharon Gilaie-Dotan, PhD UCL Institute of Cognitive Neuroscience Alexandra House 17 Queen Square London WC1N 3AR Tel: +44-20-76791122 Mob: +44-78-1450-3536 Mob: +972-54-4736181 ```
 Re: [Marsbar-users] Question about circularity From: Matthew Brett - 2013-11-17 01:17:24 ```Hi, On Fri, Nov 15, 2013 at 12:18 PM, Sharon Gilaie-Dotan wrote: > Dear Matthew, > > With respect to the question I have asked a while ago, I re-read your answer > and want to understand one more bit (if I may) - > you wrote that > "The sampled values don't depend on the design, they only use the > images and the scaling factors in the SPM design. These in turn only > depend only on the mean signal in the images, not the design. So, > yes, they should be independent." > > Can I then ask what you refer to when you write the mean signal in the > images. Does it relate to the mean signal of each participant's image (i.e. > the mean GM intensity of that image of that participant) - or perhaps the > mean signal in a specific voxel across participants? or perhaps something > else? Marsbar will open each of the images specified in the SPM design, and extract the mean signal from each image, within the ROI. The values of this signal may be scaled by scale factors that are saved in the SPM design structure (the xVol part of that structure). I think this shouldn't be the case for a VBM design - but it depends you specify it. These scalefactors apply equally to all voxels in the brain, and usually just scale the mean of the (mean voxel value within the brain - for all images) to be the same across sessions. Cheers, Matthew ```
 Re: [Marsbar-users] Question about circularity From: Sharon Gilaie-Dotan - 2013-11-17 05:35:42 Attachments: Message as HTML ```Matthew hi Thanks very much for this detailed explanation. Best wishes, Sharon ---- Sharon Gilaie-Dotan, PhD Institute of Cognitive Neuroscience & Wellcome Trust Centre for NeuroimagingUniversity College London Alexandra House 17 Queen Square London WC1N 3AR Tel: +44-20-76791122 Mob: +44-78-1450-3536 Mob: +972-54-4736181 On Nov 17, 2013 3:17 AM, "Matthew Brett" wrote: > Hi, > > On Fri, Nov 15, 2013 at 12:18 PM, Sharon Gilaie-Dotan > wrote: > > Dear Matthew, > > > > With respect to the question I have asked a while ago, I re-read your > answer > > and want to understand one more bit (if I may) - > > you wrote that > > "The sampled values don't depend on the design, they only use the > > images and the scaling factors in the SPM design. These in turn only > > depend only on the mean signal in the images, not the design. So, > > yes, they should be independent." > > > > Can I then ask what you refer to when you write the mean signal in the > > images. Does it relate to the mean signal of each participant's image > (i.e. > > the mean GM intensity of that image of that participant) - or perhaps the > > mean signal in a specific voxel across participants? or perhaps something > > else? > > Marsbar will open each of the images specified in the SPM design, and > extract the mean signal from each image, within the ROI. > > The values of this signal may be scaled by scale factors that are > saved in the SPM design structure (the xVol part of that structure). > I think this shouldn't be the case for a VBM design - but it depends > you specify it. > > These scalefactors apply equally to all voxels in the brain, and > usually just scale the mean of the (mean voxel value within the brain > - for all images) to be the same across sessions. > > Cheers, > > Matthew > > > ------------------------------------------------------------------------------ > DreamFactory - Open Source REST & JSON Services for HTML5 & Native Apps > OAuth, Users, Roles, SQL, NoSQL, BLOB Storage and External API Access > Free app hosting. Or install the open source package on any LAMP server. > Sign up and see examples for AngularJS, jQuery, Sencha Touch and Native! > http://pubads.g.doubleclick.net/gampad/clk?id=63469471&iu=/4140/ostg.clktrk > _______________________________________________ > Marsbar-users mailing list > Marsbar-users@... > https://lists.sourceforge.net/lists/listinfo/marsbar-users > ```