Re: [caret-users] Try #2
It is possible the TFCE test found no significant vertices, while the cluster method did. The TFCE generated report *does* list significant clusters near the top. (Note that the TFCE test only establishes the enhanced threshold a vertex must meet to be significant. it doesn't assign clusters as significant the way the cluster method does, but if any vertices do meet the significance threshold and they form clusters, then caret_stats writes a paint/label file and includes a list of them with areas near the top of the report.) It could look like this: ColumnThresh Num-Nodes Area Area-Corrected COG-X COG-Y COG-Z P-Value 3 0.975 1997 1286.165283 2093.53857451.174 -26.128 -1.203 3 0.975906740.481445 937.84619154.076 -30.397 17.204 3 0.975346373.274231 513.19421439.92210.387 10.669 3 0.975796432.754486 459.65432731.071 8.880 -12.815 3 0.975317145.114990 428.37191840.161 -50.331 42.466 3 0.975 13 8.440027 19.83358650.474 -57.879 33.487 If you just see the column heads, then TFCE results were negative. On Nov 22, 2013, at 4:18 PM, Eshita Shah eshs...@ucla.edu wrote: I am also wondering how I should interpret the results that are generated by the TFCE script. I know the ANOVA test on caret_command generated some text files noting nodes of significance, but from what I'm seeing, the TFCE script does not do so. How would I use the metric file to see the significant differences that were found? Thank you, Eshita On Fri, Nov 22, 2013 at 1:42 PM, Eshita Shah eshs...@ucla.edu wrote: Should I be using the Depth or the Smoothed Depth column? Are there significant differences between them? Also, I haven't been able to properly load any of the paint files onto the fiducial for viewing in caret. I'm wondering how to do that, especially if I want to highlight the areas that are found significantly different between the two groups after statistical analysis. Thank you, Eshita On Thu, Nov 21, 2013 at 4:33 PM, Donna Dierker donna.dier...@sbcglobal.net wrote: I would use the Conte69 for TFCE/cluster area computation purposes. Think of it as a neutral, unbiased atlas surface. The topology files only define neighbor relationships, so on a standard mesh, the same topo will work with a variety of configurations that are on that mesh. The ones I gave you should be fine. One thing I don't recall talking about is generating composite files of the depth metric/shape files. (Metric and shape are identical in data format. Metric was intended more for overlay/functional, while surface_shape is intended more for anatomical measures like depth, curvature, thickness, etc. But the metric menu has more features than the surface_shape menu, so I sometimes purposely use metric. For this purpose, either is fine.) The ANOVA test wants composite files for each treatment/group (maybe what you meant by factor level. So at some point you need to generate composite files to concatenate your subjects into one composite per group. http://brainmap.wustl.edu/pub/donna/FREESURFER/SCRIPTS/2009_10/SCRIPTS/gen_composite_filcav.sh login pub password download In that example, Depth was the second of multiple columns per subject. I don't recall what it is for the fs_LR stream. But if you run caret_command -metric-information on one of your surface_shape files, you'll find out which column has Depth. On Nov 21, 2013, at 4:58 PM, Eshita Shah eshs...@ucla.edu wrote: Hi Donna, Thank you for the files. I seem to be understanding so far what the sample script is doing but I do have a few questions. For the data file input into ANOVA using caret_stats, I notice it's in a different format than in caret_command ANOVA. I just want to clarify that each data file is still a metric file that contains all of the subjects for one factor level. Secondly, I realize I am using the fs_LR average open topo files you provided earlier, but for the average fiducial coordinate file, should I be also using the Conte69 average? I know you pointed out that my data is less comparable to the fs_LR standard mesh data, so I am curious as to whether I should just generate my own average fiducial file and use that instead. Let me know if I'm heading the right way. Thanks for all your help, Eshita On Tue, Nov 19, 2013 at 2:35 PM, Donna Dierker do...@brainvis.wustl.edu wrote: Here are the caret_stats and jre zip files: http://brainvis.wustl.edu/pub/donna/SCRIPTS/caret6.zip http://brainvis.wustl.edu/pub/donna/SCRIPTS/linux_java.zip login pub password download A sample script that calls TFCE is here:
Re: [caret-users] Try #2
Hi Eshita, You don't need to create an average topo of your subjects, because your data is on the 164k fs_LR standard mesh, so the open topology files in the link I provided below is all you will need to define the neighbor relationships between the vertices. You do need to make a decision or two, though: The caret_command -metric-anova-one-way feature is a valid test, but it requires a cluster-forming threshold (e.g., whatever f-stat corresponds to p=.01 or p=.025/hem). It can make a big difference which cluster-forming threshold you use, as is described here: http://www.jneurosci.org/content/suppl/2010/02/12/30.6.2268.DC1/Supplemental_Material.pdf page 6 and supplementary material figure 7 Instead, we now use Threshold-Free Cluster Enhancement (TFCE), which essentially integrates over the whole range f-stats: http://brainvis.wustl.edu/wiki/index.php/Caret:Documentation:Statistics:TFCE_Implementation Smith SM, Nichols TE., Threshold-free cluster enhancement: addressing problems of smoothing, threshold dependence and localisation in cluster inference. Neuroimage. 2009 Jan 1;44(1):83-98. PMID: 18501637 http://www.ncbi.nlm.nih.gov/pubmed/18501637 Using TFCE requires downloading caret_stats and the java runtime engine (JRE) that has been shown to work well with it. (Some JREs hang or get bogged down.) These features aren't documented in tutorials, but at least two others have managed to get it to work. If you're fine with the caret_command feature, you should be good to go. Donna On Nov 19, 2013, at 12:26 PM, Eshita Shah eshs...@ucla.edu wrote: Hi Donna, The above script was helpful, thanks. My main concern now is to run the ANOVA test (using caret_command -metric-anova-one-way). You stated earlier that I don't need to worry about the open topo file, but to input into ANOVA should I be creating an average topo file of all my subjects? Please let me know. Thank you for your patience and help. Eshita On Fri, Nov 15, 2013 at 3:50 PM, Donna Dierker donna.dier...@sbcglobal.net wrote: Scroll down eshita From: mailer-dae...@yahoo.com mailer-dae...@yahoo.com; To: donna.dier...@sbcglobal.net; Subject: Failure Notice Sent: Fri, Nov 15, 2013 10:41:08 PM Sorry, we were unable to deliver your message to the following address. caret-users@brainvis.wustl.edu: No MX or A records for brainvis.wustl.edu --- Below this line is a copy of the message. Received: from [216.39.60.175] by nm11.access.bullet.mail.gq1.yahoo.com with NNFMP; 15 Nov 2013 21:15:13 - Received: from [98.138.104.99] by tm11.access.bullet.mail.gq1.yahoo.com with NNFMP; 15 Nov 2013 21:15:12 - Received: from [127.0.0.1] by smtp119.sbc.mail.ne1.yahoo.com with NNFMP; 15 Nov 2013 21:15:12 - DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sbcglobal.net; s=s1024; t=1384550112; bh=NtxBzru9khKDAsba5SAlV4YrRgVkHeW8b9YGv0bMztA=; h=X-Yahoo-Newman-Id:X-Yahoo-Newman-Property:X-YMail-OSG:X-Yahoo-SMTP:X-Rocket-Received:Message-ID:Date:From:User-Agent:MIME-Version:To:Subject:References:In-Reply-To:Content-Type; b=WV5yG6d6wYHiVVhb2EQHyNRXQWxW1LesfyqTNaXOXmqqqOvPIFG2YS//Ij/FdfTPJj/2vVds/n4M6IksP/0A0F9p1DFQ0f99NlI5Kdnig45dD3sfU7lcXCOg4yTSnjFCUOwFbOKNDdhbE5qw7rGSY2mkoTbXduJwrvIHu6fC/LQ= X-Yahoo-Newman-Id: 757492.48204...@smtp119.sbc.mail.ne1.yahoo.com X-Yahoo-Newman-Property: ymail-3 X-YMail-OSG: gXcHylQVM1l09L.PSsdYMzRKwLwkv.NYwqaMp.6BrpjiXNE ho6UQJJXjmXuAKMCqpRBD2pMHNRD6C6IljqIBAI6R6Qm8xhM3bVHels3.Fm5 w8b7Ond.bbI2YmxgKPd57rAIo6ok.Q.vhp4ZhM8s_TaTPWlswXpD2yAlcLHq 1J3g4GvdFzgSgJ_YzgaHMEiNZaUTqjMiAsBZ30klPBT.yrdrNl9W_9TShNiA bCG4r9u36LVWGZHlQVxynSOJXA8ldy_K2eYACxDrpigxbeyqkL30yLrOmtQv rdfyk.fCTiiBI6TOCO_yOj.NPnYttzBTJEhvKwTNrhoIs3t6QxOjFKZI.zOl js7LRoCYbirS1mueqpF25Kz_lMdVsB3O6ofotbtALNwdi18tELGjU33pq6Hy .tmvnVoOH1Wy9dd9Gm1O.j2DtcMH1OWMIHROL8lAhs8hGhffYi3T7YY33LIh ujuoDMqy1hr5D4XI96NYpViITdei1lS_V51d_uov235Mw5xaWhCVuLwNwG7Q N_saCzXaf8DGq1fTdNgz2LfRRAdnh6yuEkB57kTF4BjRG2YSYgnHiPnfABCY 45DWmSS4cDvTRob2HbfUDsx5nwS0t6N4joyIUQ3I2_gz2502fqOnNC0h4mDj cZAi9sdtFy2QMAaFYDLUg62LXsFSlpCxY0gvSmu3MlUoZLuw9wFN0IDHOKBl YK9SPRb7erKBkS2zi1POQoQOpB2iyoWJsF7_XnF6H.LEnzp0BfxjMvCm0.o9 pdWNfeuoKF3UbR5wwnXXz2LY7okbApoTG7UK_gUrsqG_aLea.qRDsFi5INWP uOFJfU1pSVKYeCpn7IpEVN7BaeeVww6BI0Iwanc3H0wJtHwMm8HgCUIb4gzL S8Bls2IYSlA0OYGW_ X-Yahoo-SMTP: q5QnzDOswBAGQX7OHacFHTX9.UGNm04EhVsFT8nwx8VksFe0qzkFUA-- X-Rocket-Received: from [128.252.37.103] (donna.dierker@128.252.37.103 with ) by smtp119.sbc.mail.ne1.yahoo.com with SMTP; 15 Nov 2013 21:15:12 + UTC Message-ID: 52868ee3.3030...@sbcglobal.net Date: Fri, 15 Nov 2013 15:15:15 -0600 From: Donna Dierker donna.dier...@sbcglobal.net User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:24.0) Gecko/20100101 Thunderbird/24.1.0 MIME-Version: 1.0 To: Caret, SureFit, and SuMS software users caret-users@brainvis.wustl.edu Subject: Re: [caret-users] Different Node Numbers References:
