# The goal of this assignment is to make a SNR map from a functional dataset
#
# 1) If you want, make yourself a sub-directory in
#     /pkg/classes/psy5065/Assignments/SNR
#    and put a copy of epi_3mm.nii and epi_1p5mm.nii in there.  Or you can
#    work in SNR and just make unique output names as you run the following 
#    commands.
    
# 2) In Assignments/SNR you'll find epi_3mm.nii (fMRI data from a while back)
#    and epi_1p5mm.nii (1.5 x 1.5 x 2 mm resolution).
#    If curious, use afni to open it up and explore.

# 3) Use 3dTstat to calculate a mean image from the epi file by typing
#    the following command (this needs to be done in the directory w/ the data):
     3dTstat -prefix epi_3mm_mean.nii -mean epi_3mm.nii

# 4) Calculate a standard deviation for each voxel:
     3dTstat -prefix epi3_stdev.nii -stdev epi_3mm.nii
     # note in help file for 3dTstat (which you can get by just typing 3dTstat
     # at the command prompt) that std is computed after removing linear trend.  Smart!

# 5) Use 3dcalc to calculate the SNR for the data set (mean divided by 
#    standard deviation, through time)
     3dcalc -prefix epi_3mm_SNR.nii -a epi_3mm_mean.nii -b epi_3mm__stdev.nii -expr 'a/b'

# 6) Use afni to compare the original data with the SNR map. To do this
#    load the mean or stdev as underlay and stdev or SNR as overlay, and bond with the data.

# 7) Repeat the above exercise for the high-res data by replacing epi_3mm in
#    all of the above comands with epi_1p5mm:
     3dTstat -prefix epi_1p5mm_mean.nii epi_1p5mm.nii
     # note that taking the mean is the default behavior for 3dTstat, so you can leave it out
     3dTstat -prefix epi_1p5mm_stdev.nii -stdev epi_1p5mm.nii
     3dcalc -prefix  epi_1p5mm_SNR.nii -a epi_1p5mm_mean.nii -b epi_1p5mm_stdev.nii -expr 'a/b'

# 8) To turn in:  email caolman@umn.edu a description of the differences 
#    between the SNR maps for the low-res data (epi_3mm) and the high-res
#    data (epi_1p5mm).  Describe the signature pattern of physiological 
#    noise in the low-res data, and how you can tell that the dominant noise
#    in the high-res data is thermal.