The Utter Buffoon's Handbook on working with STIMULATE
Some notes on this manual:
Like the Afni manual also found on this page...this Stimulate manual was designed to be used while working through the program. It is in this way that a complete understanding of the program and its capabilities can be fully appreciated.
At the beginning of each section, there are some notes inRED about what the section covers. It may be that you have a certain understanding of what the section contains. If this is the case, feel free to skip ahead to a more relevant section. Sections are outlined in BLUE. AFNI buttons are presented in FUCHSIA.
Section One: Setting Up
This section describes how to load stimulate and load images.
1) STARTING UP: at the UNIX command prompt change directories into the directory that contains the images of interest. Once you are in the correct directory...type stimulateat the prompt. The stimulate icon will appear at the top of the screen...double click this icon and the program will commence.
2) VIEWING A PREVIOUSLY MADE MAP:
A) the first step is to click on theLoad. You may now choose the anatomical orientation from which to view further images. Commonly the axial orientation is chosen and click on SDT to change into the correct format. Then click Load and the slices will load. Often only the dirst slice will load and one will have to either click the up arrow to the right of Number Planes or manually enter a number...usually 10.
B) R click onLoad and then click on Map from the pop-down menu. Then choose the map of interest. If wishing to view a BoldFold map choose the BF mag map as it is most useful. Make sure the format remains sdt and then click Load.
Section Two: Altering a Map
This section describes how to change the appearance of a loaded/calculated map.
The features that determine the appearance of a map are generally located in the upper right hand corner of the Stimulate interface.
The most important of these isPos as it is by L clicking on this button that activations will show up.
After clicking Pos the two areasCeiling: and Floor: control the conditions that a given voxel must meet in order to be coloured. Increasing the floor (by clicking on the number to its right and entering a value) will generally reduce the number of voxels which will show up activated. How much activation is reduced depends on the type of map that was calculated. If a correlation map was calculated at 0.6 and you move the floor from 0.01 to 0.5 nothing will happen because no values below 0.6 will show up. Changing the floor and the floor in a % change map to 0.05 will remove activations for all voxels that did not show an intensity change of 5%.
ALWAYS REMEMBER IN STIMULATE THAT WHENEVER YOU ALTER A CRTIERIA ie Floor YOU MUST L CLICK ONUpdate Map FOR THE CHANGES TO BE SHOWN!!!
The buttonsMap Max. and Map Min in the upper right hand corner of the interface control the colour pallette that will be displayed in the voxels. There is no right or wrong range for these values. The default range is from 0.01-0.10 which gives a map that has mostly yellow activations. Increasing the max value to a number like 0.3 or so should give a map with a "hot" colour scheme similar to Afni. This is all preference so if you have values that make you smile then by all means do not feel guilty by applying them.
Section Three: Making maps with STIMULATE
This section provides detail on how to calculate the various statistical maps in stimulate's repetoire: T-test, PXC and TXC.
The first step that is needed when calculating a map is to load a stim file. This is done by R clickingLoad and L clicking stim. Next choose a stim file. Theese files are named by the task involved followed by the sdt suffix. For instance if one wanted to do map on case...pick case.sdt and make sure the format at the bottom of the window is in sdt. then click load.
There are several load options for sdt format. Important ones for making maps include First Time: the default for this is 0. In many instances when calculating a map you want to ignore the first few slices because the machine is equilibrating the hydrogen nuclei of the individual in the scanner and this is neither a control or stimulation period. Thus setting this to another number will ignore these equilibration scans. In many cases there are 5 scans so the first time should be 5. The first and last plane lines control the number of planes loaded. This is generally 10. When happy with the parameters selected hitDo It!
1) Once the stim file has been loaded you are now able to calculate a map. First, R click Calc Map. Then L click t Test.
At this time you can either load previously saved parameters (discussed below) or write a new parameter set. When loading the stim file the 1st 5 equilibration scans should have been ignored. Now one is ready to enter periods.
2) Click on either Cntl or Stim depending on what the first period is.
Now fill in the Start and End lines with relevant numbers. For instance if the first period of stim or cntl is 18 scans enter 0 in start and 17 for end. Remember so much of fMRI starts at 0 so when adding it is often neccessary to add N-1. To beat this point into your head if the first period was 10 scans start would be 0 and end 9. I am not trying to mock the readers but only protect them from botched analyses that I made for this very samw reason
Continue filling in the rest of the periods in this manner ensuring to asign each period as Stim or Cntl correctly.
When done click Plot to see a graph of the parameters. This is a good check to see that the graph looks equal...a misrepresented stim or cntl period will be very obvious in this display.
3) Next L click Add under the box headed by T Tests.
L click All which is under the box that says T Test Periods.
For those interested this runs a T test comparing the mean during stim scans to the mean during control scans.
4) BckGnd Thresh: limits evaluation to voxels that have a signal greater than the number entered in this space. By convention we choose 200.00 which removes any small activations that are outside the brain.
5) Confidence: voxels whose t-test values are below this threshold will be given a map value of zero. A very high confidencs level will work well and thus a confidence of 0.999 or greater is the norm.
6) There are two types of maps. Confidence...uses the t-test confidence values whereas Percent Change employs the ration of ave. intensity change to the mean intensity of the control period.
PZMV: period zero mean variance can be used to correct a baseline drift if suspected.
7) When all the parameters for the t-test have been entered you must save the parameters. Thus, if doing a t-test on data of the same form one is able to load these parameters without having to re- enter them.
8) Finally hit Do Map Calc and the map will calculate see section 2 for info on how to view the map if needed.
PXC: correlates data with a box car waveform.
1) R click Calc Map
2) L clickPXC
3) Ensure that the number of images loaded ignores the first few equilibration scans...usually 5.
4) EnterStim Cntl periods by typing numbers beside start and end. Periods do not have to encompass the entire time course. A smaller waveform can cover the entire time frame by shifting that wave.
5) UsePlot to view the boxcar waveform at any time. The wave can have sloping sides if you do not have consecutive periods. IE if the first ten scans are Stim and the next 10 Cntl one could for the first period enter start 0 end 7 and then start 11 end 18. The program will slop the waveform between 7 and 11 etc.
6)BckGnd Thres: see T-test section
Corr. Thres: ditto
7)Corr Start and Corr End are used to shift a waveform. Use the arrows or enter a number manually to shift a waveform so that multiple identical correlations are calculated.
NB: this option is available only if the specified waveform is of less duration than the loaded number of scans. IE if correlating a 180 scan waveform with a 180 vol. stim file shiftinf options will be denied.
8) When done entering waveform L clickSave and then save the parameters remembering that the same file can be loaded if a similar calculation is need in the future.
9)MAX: gives map of each voxels greatest correlation over all shifts
Shift: map of correlations at the maximally correlated waveform.
% change: maps the % change from Stim to Cntl
10) ClickDo map calc
TXC: period cross correlation
1) R click Calc Map
2) L clickTXC
3) R clickLoad
Txc parms...for previously saved parameters
ascii text...for manually entered waveforms from the text editor.
CREATING A WAVEFORM FROM A STIM FILE
This allows one to pick a waveform from the data with which to correlate all other voxels.
a) Load a stim file (discussed previously)
b) It helps to overlay a map (perhaps a BF map) in order to see the activated areas. This is also discussed at length earlier in the manual
c) Choose a regino of inerest (ROI) underMouse Fnc: I find that the best one for this task is Poly...that is a preference as the others will also work. If choosing poly find an area on the brain that has significant activations that you are interested in and then hold down L button on the mouse as the encircle these pixels.
d) UnderTime Course L click Full
e) L clickPlot...
f) R clickSave choose ASCII Avg TC if ROI is a box or polygon.
chooseASCII Rax TC if ROI is a point.
This is important because in order to load a waveform for a TXC the Ascii file must have only one c olumn. If you choose Ascii Raw TC for a ROI with n voxels there will n columns ans you will not be able to load it as a waveform. Another thing to keep in mind is that if you picked a ROI and then changed to a different one make sure to click Clear Roi or else the Roi will include all the Rois not ust the last one that was specified. These ascii files can be passed to AFNI for comparison by changing the suffix to 1D.
4) Once a waveform has been specified from a ROI or enter manually choose the saved waveform which will be in the ascii tc in the TXC menu. Alter the parameters as discussed previously and then save it as txc parms. All other options have been explained previously so when finished hitDo Map Calc!
Section Four: Setting Talairach Co-ordinates
This section explains how to normalize a subject's brain to Talairach space.
1) For each of the three anatomical orientation (ax, cor, sag)
a)Load the given anatomical...eg ax.sdt (discussed previously)
b) R clickSave
c) L clickAnat
d) L clickReal (4 byte float)
e) L clickSave..before saving ensure to change the name to something memorable i.e. axfloat.sdt
2) L clickTala Coord...
3) R clickLoad
4) Select all three orientations and then load the respective float file...eg axfloat.sdt.
5) If you would like a better view of the detail of the scan the magnification can be altered by adjusting the number in the lower right hand corner. Default mag. is set to256 = 2^8 and increasing this to 2^9 or even 10 will allow better detail especially helpful in picking the AC (anterior commissure) or PC.
6) Next pick out theAC and PC points in the sagittal section. To pick the AC first click AC and then find the point on the scan and L click. Next click PC then find the posterior commisure. The best sag slice for this is Sag Idx: 9.
7) In the axial slice next pick theAP, PP, RM and LM in the same fashion. Adjust magnification if needed. Remember to first click thich point you are picking ie AP and then find that point of the scan and click. One thing to keep in mind that is important is on the AP and PP ensure that they are in the midline. This means that you find the farthest anterior or posterior extent of the brain and then shift this point into the longitudinal fissure. The reason to do this is you want the axis for these two to run parrallel and through the longitudinal fissure.
8) Next surprise surprise one moves onto the coronal section. For a alisting of what the abbreviations for each point are check Stimulate Manual pg 67. Choose these points in the same fashion. First click eitherSP, IP or IC and then find the correspondeing point on the scan...shift it into the midline and then click there. This is again to ensure that the Z axis lines up with the long. fissure.
9) When all points have been set click onProj and Pts in the bottom right hand corner to check that the axes line up reasonalby and that you have not called the inferior cerebellum the antrior commisure etc.
NB although I neglected to mention it...it is imperative that one flip through the slices in the different orientations to find the highest, lowest, leftest etc. point. Not all the points in the ax or coronal orientation must come from the same slice.
10) Next clickGrid in order to see the grid on the brain
11) L clickDims... then L click apply. This normalizes the dimensions of the brain in question to the dimensions of the Talairach brain.
12) Lastly L clickSave and then name and save your co-ordinates so that they can be loaded at a later time.
I hope this short manual provided more help than headaches. Stimulate does not have the same capabilities as Afni or SPM but it is due to this fact that it is a lot easier to use. I realize this manual is basic but I hope it has provided some help in allowing you to get the results that your data are just waiting to show.
For comments or questionsfeel free to reach me at