 |
 |
Hi Marko -I downloaded TOM from your site some days ago, but I didn't try it yet because I wanted to use SPM8. Then is it possible to generate the priors with TOM and SPM5 and then use them in SPM8 (replacing the apriori folder files), or should I process all the analysis with SPM5? I have read that it's possible to generate your own template using your own T1 images, I'll try to do that. -I'll check FDR, thanks for the tip! :) -Can I ask a couple of basic questions more? Do researchers use "uncorrected" option when showing activation maps, or is it more trustworthy to use FWE or FDR? The voxel extent threshold sets the minimum size of the clusters that will be shown. Does it work only after the statistical map is calculated, merely not showing the clusters smaller than the value? Thank you very much for your help! Have a nice day :) Álvaro On Wed, 25 Nov 2009 16:55:41 +0100, Marko Wilke <[log in to unmask]> wrote: >Hi Alvaro, > >> However, I am not sure of how should I install these templates on the spm8 >> folder, or if it does actually work for the SPM8 version. > >You _could_ use the CCHMC priors for segmentation in spm5/8 but I would >advise against it. First, they are smoothed, and second, they are not >the first choice anymore. Instead... > >> How could I use CCHMC(or another) template to process pediatric functional >> data in SPM8? > >I would recommend you check out the TOM toolbox, as described in Wilke >et al, NeuroImage 2008, 41: 903-913. I think we also have a preprint on >the IRC website at https://irc.cchmc.org/software/tom.php but if you >cannot get to it, let me know. The templates generated by the toolbox >are, in my humble opinion, far superior to what we generated some years >ago based on the CCHMC dataset. We are working on a dedicated spm8 >version but for starters, the current version can be installed in spm5 >and the priors can be supplied to unified segmentation (check the >extended option where it specifies the tissue priors). > >> -I have another beginner's question in reference to the FWE threshold: >> Is the 0.05 value the commonly used, or can I change its value to discover >> truly activated areas not shown with it, knowing that the possiblity of a >> false positive increases by doing so? > >There is no law requiring p = .05; it is a customary setting that you >can deviate from, but you better have a good explanation. Alternatively, >I like to use FDR correction, which is also built-in in SPM8 but >somewhat hidden (see >https://www.jiscmail.ac.uk/cgi-bin/webadmin?A2=ind0907&L=SPM&P=R360). It >emphasizes sensitivity while FWE emphasizes specificity (in other words, >FWE is far stricter). > >> Thank you very much for all your posts and help, it's really nice to learn >> reading the list:) > >You're welcome :) And: I fully agree! >Marko >-- >===================================================================== >Marko Wilke (Dr.med./M.D.) > [log in to unmask] > >Universitäts-Kinderklinik University Children's Hospital >Abt. III (Neuropädiatrie) Dept. III (Pediatric neurology) > Hoppe-Seyler-Str. 1, D - 72076 Tübingen >Tel.: (+49) 07071 29-83416 Fax: (+49) 07071 29-5473 >=====================================================================