 |
 |
Yes, Janine, This makes sense to me - I just got this thought as well. Thank you very much, Sim Luck On Tue, Feb 28, 2017 at 1:01 PM, Janine Bijsterbosch < [log in to unmask]> wrote: > Hi Sim, > > What values to use for -thr and -uthr entirely depends on the values that > the image contains in the brain region that you want to be your ROI. > Therefore, this is something you need to decide by looking at the image you > are using to create your mask (in fslview), so there cannot be a rule of > thumb. > > Cheers, > > Janine > > ----- > Dr Janine Bijsterbosch > Postdoctoral Researcher > FMRIB Centre, University of Oxford > John Radcliffe Hospital > Oxford, United Kingdom > [log in to unmask] > > On 28 Feb 2017, at 18:22, Sim Luck <[log in to unmask]> wrote: > > Hi, Janine, > > Thanks so much. Then how to decide the -thr and -uthr parameters? Is there > any a thumb law? > > Thanks - I did not know the applywarp command in the ROI analysis. But > when I do the ROI analysis, I use the "convert_xfm", and then "flirt" > (i.e., register ROIs to subject space, and apply the transformation matrix > to the ROI), followed by "fslstats" > > Thank you very much, > Sim Luck > > > > On Tue, Feb 28, 2017 at 12:06 PM, Janine Bijsterbosch < > [log in to unmask]> wrote: > >> Hi Sim, >> >> When deciding which commands you need to use to make an ROI it is import >> to understand what type of image you are working with and in what space it >> is. For example, if you look at the Harvard-Oxford atlas image you can see >> that it contains values everywhere in the brain, and therefore to isolate >> the brain region you are interested in you need to apply a threshold (to >> remove everything below a value using -thr and/or to remove everything >> above a value using -uthr). It is also important to make sure that the >> resulting ROI image only contains 0s outsize the ROI and 1s inside the ROI, >> which is achieved by binarising using -bin. Lastly, the ROI image and the >> EPI image need to be in the same space before extracting the BOLD >> timeseries, so you may need to use the applywarp command (followed by >> binarisation) to make sure your ROI image is in the right space. >> >> You can find out more information about all of these by looking at the >> FSL wiki and usage information for commands. >> >> Best wishes, >> >> Janine >> >> >> >> On 28 Feb 2017, at 17:51, Sim Luck <[log in to unmask]> wrote: >> >> A further question: if I get a mask from others (it is not necessarily >> made using FSL), should I also use the command fslmaths, -thr, -bin to >> process it first before I use it? >> >> Thank you very much, >> Sim Luck >> >> >> >> On Tue, Feb 28, 2017 at 11:25 AM, Sim Luck <[log in to unmask]> wrote: >> >>> Hi, List experts, >>> >>> I am still not very clear with the option "-thr 10 -bin" in the ROI >>> making process. For instance, the right amygdala (top plot) was made >>> directly using FSL atlas database. The bottom part was made based on the >>> above one, but using the command, fslmaths Left_Amygdala -thr 10 -bin >>> Left_Amygdala_thr10. Which one should I use? >>> >>> When should we use both "-thr" and "-uthr", and how to choose the >>> parameters? In some examples, people use -thr 2, -uthr 2; while someone use >>> -thr 6, -uthr 6, and others use -thr 10, -uthr 10. I am not clear on how >>> should I use them, or just use one, say "-thr". What is the thumb law? >>> >>> Is that not good if we do not use "-bin"? If I do not use it, what is >>> the influence? >>> >>> >>> >>> >>> Thank you very much, >>> Sim Luck >>> >> >> >> ----- >> Dr Janine Bijsterbosch >> Postdoctoral Researcher >> FMRIB Centre, University of Oxford >> John Radcliffe Hospital >> Oxford, United Kingdom >> [log in to unmask] >> >> >