HFM Version 1.0

• Put in the loadret directory copies of the following:
• Raw structurals E2097S7I*.MR
• Coplanar Images: HC-Subj+8+2+1_001.img
• ConversionDataKS.mat from segmentation procdure (this is the cropping information)
• run1_p.mat, run2_p.mat, Cproto files converted to matlab file format for all the paradigms
• AIRed timeseries S1.run1.bshort - need not be in same directory, but know where they are

• 1) Enter parameters, import structural EPI data

% matlab -nojvm

>> HFM

Number of anatomy inplanes: 11 (# of slices in the volume, before interpolation)
Number of experiments (P files): 1 (This is the number of runs)
Number of temporal frames (functional images per plane): 116 (Enter the number of images in the longest run)
Number of temporal cycles per experiment: 5 (Any random number is fine here)
Discard the first (n) temporal frames: n= 0
Directory of anatomy images is /data/subject/loadret
Press <return> to accept, or enter a new directory: (enter)
Size of epi anatomies? [128 128] (EPI resolution)
Size of signa anatomies? [512 512] (Structural resolution)
Size of functionals? [128 128] (EPI resolution)
Target size? [512 512] (Destination resolution = structural resolution)

Fake anatomies from functional data? 1 = yes 0 (hit 1 for yes if you did not acquire an extra set of coplanar EPIs)

Give me offset x: epiplane + x = functional plane (0) 2 (This would be for slices 3-13)

Enter EPI Coplanar Image Name: HC-Subj+8+2+1_001.img

ans =

A1.s1.bshort

Enter correct series number: [default 1]:1

• 2) Import high-resolution structural spin-echo images, and align with EPI structural images

Select a good slice near the center of the brain. Then, HFM -> Signa Anatomies

>> Give me offset x: epiplane + x = functional plane (0) 2

offset =

2

ans =

E1636S3I11.MR

Signa exam number and series number ([1636 3]? [1636 3] (or hit enter if the default is correct)

Enter shift if known (eg. [-8 -4]), or hit enter to continue and find the shift: (enter)

Clicking on 5 homologous points in the structural and functional images.

shift =

-8 1

Hit 0 to try again, 1 to accept: 1 (Do several trials until you get a number (xshift, yshift) that does not vary by more than 1.Write the final number down somewhere)
 

• Switch slices ... this will update the images to the hires ones.
• Close Figure 2.
• 3) Crop Anatomies

HFM ->Clip Anatomies

• Hit Y to auto-extract crop from file conversionDataKS.mat. This is best.
• Hit N to crop yourself. Select the region - be sure to include all the gray matter. In general I choose superior to the HC in the most anterior slice, and inferior to the FG in the most posterior slice, and laterally the edges of the brain.
• 4) Import EPI Time series
  

HFM ->EPI -> AIR tSeries

• Enter the pathname and directory for each timeseries, ie, each experiment in the run:
• Eg. For /space/zeineh/subject/run1/run1.s001.bshort, type
Directory: /data/subject/air/run1/

Filename: run1

• Then, if slices 3-13 were selected, type 3:13
• It will work for about an hour or so as it imports the functional data and interpolates up the functional data to match the resolution of the structural data. I default it to nearest neighbor interpolation for honesty's sake. If your functionals are not an even multiple of your structurals, I suggest changing this hard-coded NN interpolation to linear, bilinear, or bicubic in the file HC_AIRTseries.m in the HFM-Code directory.

• 5) Register time series with structurals, and import unfolding transformation
  

Special -> mrAlign

• Enter in parameters

>> Current list of subjects:

HC-PinkFloyd/ HC-Rush/ HC-Yes/ HC-PamAnd/ HC-Subj/ HH-Farihi/ HH-Rex/

Subject name: HH-Subj
Enter size of inplane anatomy pixels/mm in x,y and z directions
Default is [256/260,256/260,1/4]: [512/200 512/200 1/3] (this is the size of your anatomic volume before interpolation #pixels per mm)
Finished saving AlignParams.
/usr/local/mri/anatomy/UnfoldParams.mat not found. Please enter values.

Enter size of volume anatomy pixels/mm in x,y and z directions
Default is [256/240,256/240,1]: [512/200 512/200 7/3] (this is the size of your anatomic volume after interpolation #pixels per mm)

• Then go to HFM -> EPI -> Compute Trans
• Verify rotation by selecting various planes in figure 1, and hitting Select -> Check Rotation in figure 2.
• Close figures 2-5

• 6) Create flattened anatomy maps

Init -> Full Analysis -> Create Flat Anat

• Hit b for both hemispheres
• Just keep hitting enter until it is done.
• Close Fig 2

• 7) Manually adjust coregistration between EPI time series and high-res structurals

HFM -> EPI -> Shift tSeries

• Click Generate Selpts, then Save Selpts
• Use arrow buttons to shift image until overlap is good.
• Check different slices by clicking on slice # in figure 1, and then clicking update to view overlay on EPI
• You want to check slices throughout, that is, 2-10. You don't need to check every single slice, though. Slices 2,5,10 should do it.
• Click execute to shift

• 8) Mask all zero points in data
  

HFM -> EPI -> strictmean  (This routine masks the data, so if any pixels are zero at any timepoint, then that voxel will be zero for the entire time series. If you have more than one run, and if some runs have more or less time points than other runs, you want to tell strictmean the exact number of time points in each run, because it zero-fills all of the empty time points on the shorter runs, and strictmean will then zero out the entire run!)

• Number of images in run 1: [116] (enter if correct)

HFM -> EPI -> remake firstslice  (This will recreate a volume image of the 1st time point)

• 9) Project 3-D volume EPI time series to flat space, and create (empty) correlation map files in flat space, and correct gray scale scheme
  

Init -> Full Analysis -> Create Flat Timeseries, Create Flat Correlations
HFM -> Fix Flat Anats

Note that these correlation maps only correlate with a sinusoid with a frequency of the number of cycles you specified earlier. It is much easier and more precise to do the next step instead.

• 10) Create correlation maps in 3-D and 2-D that correlate with your paradigm files: Note you need paradigm files for this step. 

HFM -> Statistics -> Inplane Correlations
HFM -> Statistics -> Flat Correlations

• 11) To visualize activity, click L or R, and View-> Phase Map. Hit HFM -> colorph to make it look nice, and adjust the threshold bar at the top.
• 12) You will want to add the demarcations ... see the next section for how to do this.
• 13) After you are done with demarcations bilaterally, if you hit Fill Regions on the demarcation window, it will create ROIs for each of the 7 subregions on each hemisphere (the rigth column of the demarcation window). You can plot timeseries from each subregion, or extract all timeseries to a spreadsheet.

To create the regions:
(Demarcation Menu) Fill Regions

To plot a single region:
(Demarcation Menu) CA23DG (or any other region you want to plot)
(Figure 1 Image Window) Load tSeries
(matlab window) a=tSeries(:,selpts(1,:)); b=mean(a,2); figure(3); plot(b);

To store the time series of all regions:
(Demarcation Menu) Region TS
threshold images (1=yes, 0=no) [0]  0  (better to not threshold to make analyses unbiased)
Paradigm Drift Correction Value (ie baseline in paradigm file) [0]  0  (ie which values in your paradigm file correspond to a physiological baseline for fMRI signal intensity)
Drift Correction Order [1] 1 (1=1st order, 2=2nd order, 0=0th order: what happens is that a polynomial of this order is fit to all of the timepoints who are baseline timepoints as specified above. The data is divided by this polynomial curve to deliver drift corrected % change from baseline signal intensity).
Saved RegionTS TS TS_points
Saving RegionTS.txt (you can view this file with a spreadsheet program)

If you want to process another run on a subject who was already unfolded and processed, you can use the following shortcut:

• Copy the following files from that subject's prior loadret data directory:

mkdir /data/HC-Subject/loadret2
cd /data/HC-Subject/loadret2
set subject = /data/HC-Subject/loadret
cp ${subject}/AlignParams.mat .
cp ${subject}/ExpParams.mat  .
cp ${subject}/Fanat_left.mat  .
cp ${subject}/Fanat_right.mat  .
cp ${subject}/anat.mat  .
cp ${subject}/mrLoadRetPrefs.mat  .
cp ${subject}/roi14.mat  .                        (Note, get the latest ROI file)
cp ${subject}/HC.mat  .

 

• run MATLAB - nojvm, and run the script changeExpParams

>> HC_ChangeExpParams
# of experiments [2] 2
Images per experiment:  [96] 103
   
• clear all, then run the HFM module

>> clear all
>> HFM
HFM Version 1.0 by Michael Zeineh
mrLoadRet, version 1.0. Matlab 5.0
Initializing general workspace variables
Initializing Callbacks
Checking anat file
Loading anat.mat.
Reading mrLoadRet Preferences file.
Done loading mrLoadRet-1.0
 
• Import the EPI fMRI data

HFM ->EPI -> AIR tSeries

• Enter the pathname and directory for each timeseries:
• Eg. For /space/zeineh/subject/run2/S1.scan2.bshort, type
Directory: /space/zeineh/subject/run2/

Filename: scan2

• Then, if slices 3-13 were selected, type 3:13
• It will work for about an hour or so
• Align the data

HFM -> EPI -> Shift tSeries

• Click Generate Selpts, then Save Selpts
• Use arrow buttons to shift image until overlap is good
• Check different slices by clicking on slice # in figure 1, and then clicking update to view overlay on EPI
• Click execute to shift
• Mask the data, and remake an image of the 1st volume timepoint
  

HFM -> EPI -> strictmean

• Number of images in run 1: [116] (enter if correct)
• Number of images in run 2: [116] (enter if correct)

HFM -> EPI -> remake firstslice

• Create the flat anatomy and correlation map files

Init -> Full Analysis -> Create Flat Timeseries, Create Flat Correlations

• Create a correlation map, putting in this directory the appropriate paradigm files: run1_p.mat, etc.

HFM -> Statistics -> Inplane Correlations
HFM -> Statistics -> Flat Correlations

Creating a paradigm file:

Back to main