## Master's Thesis Research Project
This is my research project from my Master's Program at Virginia Tech, see here:

ftp://bsl.ece.vt.edu/static/ReproducibleResearch/ISBI11-BOLD/BOLD-ISBI2011-preprint.pdf
http://scholar.lib.vt.edu/theses/available/etd-09212010-215625/unrestricted/Chambers_MicahC_T_2010.pdf

This is not actively being developed, it is more here for archival purposes, or in case someone wants to grab a concrete version of the BOLD model.


############################
To generate a parameter map:
############################

#create a file like this:

#Begin File "params" #
#          Tau0          alpha             E0             V0           TauS           TauF        Epsilon 
        1.48256       0.280681        0.35817      0.0262547        1.84741        3.18442       0.003857 3.4 1
        1.20253       0.309557       0.326097       0.035857        2.16765        2.27181       0.085933 3.4 1
       1.950512        0.31665       0.348487     0.02661222         1.6572        3.71875       0.756764 3.4 1
       1.454322       0.321322       0.369314       0.035621        0.99371        2.77384        1.34784 3.4 1
        1.15114       0.352777        0.38027      0.0262798        1.98008        2.33257        1.64476 3.4 1
#End File "params"  #

#then run:
$ brainid/build/code/parammap_generate parammap.nii.gz  params -t template.nii.gz 

#template.nii.gz can be any 4D image, whose size/shape/orientation you want to
#copy. I am not sure it will run without one. -T and -s are used to generate 
#regions of size/and roughness, you will have to play with them to get regions

#If you have a mask image, and want to turn that into regions pass the mask instead 
#use the follow. This is the preferred method. I used the mask image provided by
#Possum "./fsl/data/possum/activation3D.nii.gz" 

$ parammap_generate parammap.nii.gz  params -m mask.nii.gz

#To used Possum the custom version in the repository will have to be used:
#https://dmz-server/svn/POSSUM-mod/


#email me about this if you want to use POSSUM because it is sort of tricky


################################################################
To generate active timeseries (without POSSUM or parameter maps)
################################################################

$ brainid/build/code/boldgen -bf bold.nii.gz -ot 2.1 -st .0005 -n 1 -sf truestate.nii.gz -randstim -rstimt .1 -rstimp .1 -snr .001 -drift .0005 -X0file params -rstimfile stim

#this will generate a single voxel, 2.1 TR image with the parameters from params (note that it is slightly different from the above
#if you give more than one section (-n 2) then it will read in the second row of the params file and so on....
#Note that -snr isn't really the snr, it is just the standard deviation of the noise. -drift is the standard deviation
#of the drift steps. If for some reason this doesn't complete you may need to decrease -st to .0001.

#          Tau0          alpha             E0             V0           TauS           TauF        Epsilon V Q S F
#Begin File "params" #
        1.48256       0.280681        0.35817      0.0262547        1.84741        3.18442       0.003857 1 1 0 1 
        1.20253       0.309557       0.326097       0.035857        2.16765        2.27181       0.085933 1 1 0 1 
       1.950512        0.31665       0.348487     0.02661222         1.6572        3.71875       0.756764 1 1 0 1 
       1.454322       0.321322       0.369314       0.035621        0.99371        2.77384        1.34784 1 1 0 1 
        1.15114       0.352777        0.38027      0.0262798        1.98008        2.33257        1.64476 1 1 0 1 
#End File "params" #

#bold.nii.gz - the noise free BOLD signal
#drift-bold.nii.gz - the BOLD signal with just drift
#noise-bold.nii.gz - the BOLD signal with drift and white noise
#carrier-bold      - the signal you want to input to calc_parammap
#
#truestate.nii.g   - the true state parameters changing in time
#
#stim              - contains the activation level time pairs

################################################################
To use calc_parammap for the above case
################################################################
$ mkdir output
$ brainid/build/code/parammap_calc  carrier-bold.nii.gz  -l 0 -S 0 -t 2.1  -p 1000 -d 1500 -s stim -f 0 -w 0 -C .005   -o output/ -e 0 -D 0 

#this will use 1000 particles (-p), euler integrate with step size of 0.00066666 (-d),
#will not flatten the prior (-f), won't use delta (-l) or "smart knots" which don't work (-S). 
#the stimulus file (-s) is from the output of boldgen, and should match whatever the input BOLD
#sequence used. The weight function is open to change (-w), as is the weight scale (-C) 

#histogram output will be put in output/ (-D 0)
#and no time points were erased form the from (-e 0)

###############################################################
To analyze histogram output
###############################################################

$ cp stim output/
$ brainid/scripts/makehist.py output/

#give it a little time

###############################################################
To evaluate error etc
###############################################################

$ cp stim output/
$ brainid/build/code/parammap_eval output/

#if you want the true error, as opposed to the residual, you have to create a parammap for the 
#parameters and then copy that (possible single voxel) image to output:

$ brainid/build/code/parammap_generate parammap.nii.gz  params 
$ cp parammap.nii.gz output/
$ brainid/build/code/parammap_eval output/