fMRI brainvoyager (2).docx

C/agvirreIndirect measure：We are not directly measure the neural activity, but the fmri signal (the blood flow). So the fmri signal does have direct absolute interpretation, must be compared between states studies close together in time.(dental work, breathing) make two groups differences. Blood flow response can be influenced by many factors. Make a claims from one study you have done.A generalizable result: One type of group, but be caution to generalize into different groups (undergraduate students & lawyers & teachers & children). extended to other groups.How can go wrong: alter blood flow(drgus, hormonal states, age, gender). Sometimes there are some neural activity differences in different conditions, perhaps there are some differences in neural responeses but it may not be measured by firm. One way that can happy is that it is not a bulk change or stable change. Since fmri show an average response of the time space.Voxel, 60,000 perhaps be an image. How to make a claim from a well make brain to mental thoughts of a man. the brain area for “l(fā)ove”. How to make inference mental operation. Brain activity and behaviour: Forward inference : determine which brain region is associated with an isolated behaviour. Parametric study : Focal Reverse inference: use local brain activity to (determine which mental states are evoked by a complex behaviour) identify mental operation. No other brain area activated. Just for pain. Good especifition. Are you lying? Mind reading is possible? Look differences between Multi-voxel classification: use distributed patterns of brain activity to predict which mental state is being experienced. Signal and noiseFourier transform: analyse signals, power spectrum (功率譜)。 Only changes that are quite slowly can be detectable by the fmri, changes that quite rapidly ups and down can not detected by the fmri (not pass well). Like every 3/4 seconds. Design experiments that were really changes slowly. 5mins faces, 5mins houses would be best, but that is not the whole story because of the noise. Because of the stuck between the hemodynamic (人的血液動力反應) response suggest the low frequency of the experiment or stimulus, while intrinsic noise（固有噪聲）suggest the high frequency of the stimulus. So how to balance between these two. 20s would be a better choice. Block design. 20s-30s seconds. Frequency repretations. What is the noise? could be heart rate, breathing, brain areas/cortex communicated /connected with each other. Frequencies.Experimental design has low frequency. Smooth the data . we normally smooth 6mm. howe ver, there are other people who smooth data in 8-12mm. do not use GLM, if you do it, you want to .3*3*3 brain module. Your anotomical data is high revolution than your function data. . fMRI: lectures from Frank:Presentation: tell the presentation to start-presentation starttell the scan start.What the final analysis are going to be. Planned everything, know in advance what analysis software you are Gonne use. Fmri safety course, you should take the safety course, they will show all the worst thing that could go wrong. Video frequency in the scanner, that is quite dangerous. Minimize the bad things. Projector, high resolution, eye trakler. Programme to run the computer: Computer will control the delivery of the stimuli, potential programes that could run the computer is : presentation (which is the one we generally use), eprime, and matlab, each one of those is ok. Show things into the scanner, need the PC and scan talk to each other. Once we go into the brainvoyager, brianvoyager always give an option to skip several volumes. It is good to get rid it. The reason is : If you look at the statistic analysis, all the data that incluse some conditions that sometimes goes to your baseline, and if your baseline includes these noise and it annoys your baseline, so just skip some volumes that does not contribute at all. But it is important to keep track of the data if multi people work on the same data.Presentation log file: it is good to have the log file from the stimulus delivery. So the computers is showing the stimulu, it is really handy if keeping track of the log of what is going on. When it is showing, what is up the scanning and what is hearing from the scanner. You need have the log file from the scanning session. if you have the log file and have it logged : log the response , log in all information, you can go back and check everything find everything(we did not log one of our participants responses, and then we want to analyse, but we did not have these responses since we did not log in), one word, if you do not have these stuff at the beginning, then you do not have these at all. Two ways to do timing with brainvoyager: uniform time,2000ms TR=2. If you want to see the brain in ms, you really need to do it by yourself, otherwise, the scanner just take the brain every 2s.When the subject was in the scanner, what happened to the scanner or computer. At the end of the session or at the beginning (it is up to you), we require the anatomic data. It will take 4-8mins to collect the anatomical data, the differences between 4-8mins is the longer time, the higher contrast between the white matter and the grey matter. In Friday, we did 2 functional runs and 1 aunotomic run. Every time, there is a run always produce one file, but the functional runs always produce 2 files, one is raw data, the other is motion corrected (moco) data. We general use the raw data (franks prefers to use the raw data, but in some other manuals, some of them use the motion corrected data, which is the second one of each run ). The process: (these steps are ubtertwined)Planning, stimulus presentation, measurement, pre-processing and statistic analysis Synchronisations : scanner waits for 7s, until signals to be stable, press button of stimulus computer and scanner at the same time. So skip first 1 or 2 volumes this synchronisation issue need to be decided since later in preprocessing BVQx will ask you if you want to skip. Log file : format is text, every 2 seconds.# Anatomical run: 192 sclies, 1*1*1 mm (the whole head)Functional run: 32 sclices, 3*3*3 mm and a whole brain volume acquires every 2 seconds.Preprocessing: Anatomical preprocessing : talariaki the brain , filter the brain in the same way (covert the brain to the talariake space) Functional preprocessing: CorregistrationCreat PRT file when you want to contrast different conditions.Brain normalization is to put brain in the talariaki space. The contrast of the brain activity when you see scrambled person and intact person.We use the wizard to put in the data. 32slices, if you have some area that you really interested in in particular, you may change the functional run to the specific brain, but we would be a little bit open to see the whole brain, to see what other brain areas may be involved. We try to measure the whole brain. Typically, you can do 32 sclices, 3*3*3 voxels and acquire the whole brain every 2 sceonds. If people have small brains, that is enough, but if people have larger brain, maybe there is something has missed. You have the option of either missing the sop superior frontal corxet on the top or the temporal cortex. All the anotomic data, everyone almost get the same thing. No matter the brain size. Where these kinds of issues become important , you want to see what happens on average about 20 brains , here you need to use the normalization . that all 20 brains at somehow should be aligned . there are techniques to allow you .present one brain at a time, you give preliminary the region you are interested. .But we usually do is to average all the brain together, then somehow you need a space that combine all these brains. All the brains have different shapes and size, even we talariki these correctly, cortex based alignment . .there is still a little chance to have brains . Create the VMR anatomical data. Create the VMR in the step 9, page 41, check the number of sclices(192)-next, check the resolution of anatomical scan/images (256*256). Then check these slices and resolutions. -check the contrast and brightness (generally the default is ok, so keep the options remain there and then press OK). After creating the VMR, you can just change the intensity of the brain. Which is inhomogeneity correction, V16 Tools. Original VMR volumesinhomogeneity correction-load v6just click the V16 file (we can also do this step before talairach).Create the FMR, Create name the first vmr (the raw data) as R1(here, it is important to use the _ underscore to avoid the confusion of file names), choose the number of volumes is 243, (keep everything as default). And then create FMR R2, here the original design for the experiment is 243 (the first one -first), but frances always run the scan longer than the experiment design, and then just press stop when it is ok. In order to be equal as to our design, let us go for 243. Then for number of volumes , let us go for 243. Then , we still not quite trust the first two volumes, so we go skip the first two volumes. To see the time course movie: go to the optionstime course movies (give the qualities of the movie).Attach PRT file: open the fmr we just created, go to FMR properties, choose the fmr run 1choose the PRT file (choose the run 1 PRT file) closeThen open the fmr run 2, go to FMR properties, choose the fmr run 2choose the PRT file (choose the run 2 PRT file) closeThus, our PRT file is attached to out fmr.Preprocessing:Preprocessing the functional data, close other file and only keep the fmr open, go to optionspreprocessing Fmriclick preprocessing slice scan time correction, 3D motion correction spatial smoothing -temporal filtering ( the mean intensity adjustment, while is not need in most cases, so dont choose this) -spatial smoothing is the one general can be down afterward, so we can leave it at this step, but we choose spatial smoothing at this time because frank said if we want use this data for group analysis, we can do the spatial smoothing). In the advanced options, we can see more details of the preprocessing. In this advanced option, I would recommend : in the slice scan time correction, I prefer to choose the sinc since we can get the better result even it is slow; in the 3D motion correction, I prefer to choose since interpolation (very slow), then we choose option, because we need the numbers for the intra-session alignment, the last volume of the fmr is more close to the VMR, so we choose the last volume as 159/159, so in the preprocessing session, we preprocessing the reference run first, and align other runs to the reference run, so in this session, we preprocessing run 2 first, choose reference volume 159/159go. After we preprocessing run 2, then we do run 1, also choose reference volume 159/159, click use other FMR for intra-session alignment , then attach the run 2 preprocessed fmr; in the Spatial smoothing, we change the FWHM to 6mm; temporal filtering ,keep it as the default. So we keep the thing remain except spatial smoothing, because we skip click spatial smoothing option, so leave spatial smoothing this time , here we do not keep it as high pass (GLM-Fourier) since if we click this , we can not do the spatial smoothing afterward. So here we choose linear trend, then choose 0.008 points or 120 secs. (brainvoyager knows TR is 2 s).Coregistration of functional and anatomical data:First step is to open the VMR (we created in the begining). volumes -3D volumes-full dialoguecoregsistration-select FMR(because we already have the VMR file open)choose the preprocessed fmraligngo Check the coregistration (align-initial alignment which you can always take what brainvoyager gives you, but as the fine tuning alignment, you can make some minor change by yourself). Important thing: when do the coregistration, FMR-VMR (choose the file that has been preprocessed), but use linked AMR-RunIAthen again click alignfine alignment(do not need to link AMR since this has been down in first step, so brainvoyager will remember it)Talairach anatomical file and functional file.Open the IIHC vmr. VolumetalairachAC-PC transformationmanualfind AC pointafter find the AC, -find the PCfind AC-PC plane-transform to ACPC(the transformation was applied automatically and that the brain is located in the AC-PC plane, this can be seen clearly in the TRA view, which shows both the AC and PC point. ). then set the other 6 pointsafter set the 6pointssay save to talACPC to trl Manually find the AC, PC, use the key ctrol T to enlarge the AC, and then ac-pc plane. Save to trl. Then automatically find the other 6 points, if there is some change, just change it. Then save trl, then ACPC to trl. So far we have almost all the file that we need to go forward.Next thing is to create VTCs.(we will learn how to transform our functional data into talairach space. The result of this process is a VTC file =volume time course) we align the second fmri to anatomical data. In this step, we need to use several information that we have produced during previous steps, namely the intra session anatomical and functional coregistration transformation matrix, the intra-session to extra-session 3D-3D alignment matrix, and the determined tralairach reference points. Here, we skip step12, and go to step 13. The reason we skip step 12 is that, 12 is usually done when we bring people in to scanners in different 5 days, then we have different anatomical data is the same but the position of the scanner is different, so you need align the anatomical data between the different days, but we just have one day, so we think we can go forward without step12. Open the IIHC tal vmr. -analysis create 3D VTC data. We need to do it twice. (once for each fmri). So we first browe R 1( preprocessing fmr), then in the top, we choose spatial transformation of functional data to TAL, then click Auto-Fill, then the IA.TRF AND FA. Trf is attached automatically, then choose, ACPC.TRF, IIHC TAL, then resulting TAL VTC file , then go. IA trf file is the file that brainvoyager create automatically. we need to align our anatomical data to all runs. so in the create VTC file, we browse run 1 and run 2. after you browse run 1, choose the options, see if there is anything to choose.and then go to analysis, click Link 3D volume time course (VTC)-browes R1.keep the file open, then we can go back and check the VTC properties, so go to the file-VTC properties, then we can know that this is R1 that has been attached.see the VTC properties, we can find the second step is linked protocal file, so then the next step we can do is to create the protocal file. (PRT file). #Create PTR file.(after preprocessed FMR, we can create the protocol file)choose the preprocessed FMR file,Then open an R1 preprocessed FMR file-file-mri properties -referenced protocal file-choose the one we have down in the last step. then go to Analysis-GLM single study- Stimuliation protocal, then name the experiment (like HWK08 ).add (independent factor, in this experiment one of the independent factor is biological motion-intact and the other is scramble), so first add -name biological -choose the color -show plot- right click the first independent factor-click HRF, then click add pred-right click the second independent factor-click HRF-go-Then restart the R2 and do it again-save PRT to R2-PRT.then open an R2 preprocessed FMR file-file-mri properties-referenced protocal file-choose the send one we have down in the last step-.then open an R2 preprocessed FMR file-file-mri properties-referenced protocal file-choose the send one we have down in the last step-.after this, we want combine them- -TAL vmr-go to anaylsis(because this is the simple study, we just have two factors)-linked 3d time course VTC filebrowse choose the first run file (which is ended in vtc file).-ok, thengo to analysisGLM single study, single subjectadd prediction, gogo to analysioverlay volume mapsmap optionschoose bonferroni p value:0.05-apply We run 2 runsthen go to analysis-GLM multi study multi subject-add to listchoose vct file-chooses sdm file, then choose every run of each subjectsave. -go.Go analysisoverlay volume maps ma optionschoose bonferroni p value:0.05-applyafter this, we want combine them-so open a processed VMR file-TAL fmr-go to anaylsis(because this is the simple study, we just have two factors)-linked 3d time course VTC filebrowse choose the first run file (which is ended in vtc file).-ok, then go to analysis-GLM multi study multi subject-add to listchoose vct file-chooses sdm file, then choose every run of each subjectsave. -go.(load MDM file), mult study multi subject GLM, we are not sure which to choose among (RFX GLM, %-transform etc.) choose the correct serial correlation, % z transform (), separate study predictors.Options-Excude voxel time courses: intensity threshold :100-goWhich gave us the result, but from the result is a little bit out of the brainBut we have not seen the contrast, so we run the Overlay ginear model contrast, -optiongoGo to analysis-Overlay volume mapsConvert map to VOI, 300Statistical data analysis : Other brainvoyager materials: Learning the vocabulary: MRI: no temporal information. / high resolution, images brain anatomyFMRI: images brain function/ low resolution, high temporal information, collect brain images every 2 seconds and last 5-6 mins, if there is a task, that would be 20-40 mins. Interpolate : interpolate英ntplet vt. 篡改；插入新語句vi. 插入；篡改.Slice time correction: if slic