From dgold@bic.mni.mcgill.ca  Fri Jun  1 17:10:25 2001
Modified by Kate Hanratty, Friday, September 20, 2002.

Recipes for processing T1 weighted MRI volumes.

--

# 1. For a linear transformation from Native --> MNI_Tal space:

batch -Q medium -H bullcalf -S -k -o '1Linear_reg.log mritotal -model
icbm_avg_152_t1_tal_lin_symmetric subject_native_mri.mnc
subject_t1tal_lin.xfm' 

# 2. Evaluate the contribution of nonuniformity of the magnetic field
  in the MRI volume

batch -Q medium -H bullcalf -S -k -o 2Evaluate.log 'nu_estimate
subject_native_mri.mnc subject_native_field_map.imp'

# 3. Correct the field inhomogeneities in the MRI volume 

batch -Q medium -H bullcalf -S -k -o 3Apply.log 'nu_evaluate
subject_native_mri.mnc -mapping subject_native_field_map.imp
subject_native_mri_nuc.mnc' 

# 4. For a non-linear transformation from Native --> MNI_Tal space:

batch -Q medium -H bullcalf -S -k -o 4Nonlin_reg.log 'mritotal -model
icbm_avg_152_t1_tal_lin_symmetric -nonlin native_mri_nuc.mnc
-transformation subject_t1tal_lin.xfm subject_t1tal_nl.xfm'

# 5. To generate a minc volume in MNI_Tal space

batch -Q medium -H bullcalf -S -k -o 5Resample.log 'mincresample -like
/avgbrain/brain/images/icbm_template_1.00mm.mnc
subject_native_mri_nuc.mnc -transformation subject_t1tal_lin.xfm
subject_stx_mri_nuc.mnc' 

# 6. To classify tissue in T1 weighted volume as grey matter, white
# matter or csf

batch -Q medium -H bullcalf -S -k -o 6Classify.log 'classify_clean
stx_mri_nuc.mnc stx_classes.mnc'

# 7. To identify structures in the T1 weighted volume:

batch -Q medium -H bullcalf -S -k -o 7Segment.log 'stx_segment
-symmetric_atlas subject_t1tal_nl.xfm subject_t1tal_lin.xfm
subject_classified.mnc subject_stx_labels.mnc'

--
Extras:

# To generate a label file with only a few structures specified

minclookup -discrete -lut_string '1 5; 3 7; 9 2' subject_classified.mnc
   new_label_file.mnc

{If you want to use the same colors, you'll say for instance '1 1; 7 7;
etc', but if you want to use other colors, you can say eg. '1 15; 46
7; etc'.}

The label numbers can be found by looking at Jakob's brain.

MRI Volume lives in:

/data/avgbrain/avgbrain1/brain/images/himaster_stereo_jacob.mnc

Labels:

  All labelled structures in one file:

  /data/icbm/cryo2/noor/Jakob-2002/jakob-temp.mnc

   (put the arrow cursor over a colour label and look in the lower
   left hand corner of the Display main window to see what numerical
   value this label has.  Look for "Lb ###")

  Individual labels:

  /data/icbm/cryo2/noor/Jakob-2002/Extracted-labels/label_###.mnc

   Where ### represents the number you looked up in the jakob-tmp.mnc
   label file...

# 36 = Right hippocampus; 101 = Left hippocampus


# the command for the cortical surface removal is a two-step
# process. First, you have to render the surface of the brain you are
# interested in: 

# Compute the threshold: 

/nil/david/public_bin/compute_cortex_threshold.pl brain.mnc

{It would seem that the number spit out by this process is too high
 for the automatic registration process. 

 Taking 80% of that number seems just right.
 Taking 70% might be better. Yupf, much better. Oh yeah.}

# To make the cortical surface object:

batch -Q medium -H bullcalf -S -k -o adelson.log/
/home/bic/david/public_bin/cortical_surface.pl scan_mri_tal.mnc
adelson_75.obj 75 

#With that, you call the program 

surface_mask2 classified_image.mnc object_file.obj
classified_image_surface_corrected.mnc  

# After that, you can use that for multiplication with your label
file. That is a three step process.

# First, match the two files

mincresample label_file.mnc label_file_rszd.mnc -clobber -like
classified_image_surface_corrected.mnc -nearest  

# Then, multiply them

mincmath label_file_rszd.mnc classified_image_surface_corrected.mnc
label_file_surface_rmvd.mnc -clobber -mult 

# Last, correct the image range

mincreshape label_file_surface_rmvd.mnc
label_file_surface_rmvd_corr.mnc -clobber -image_range 0 255
-valid_range 0 255  

# All set!

#Finally, change the labels back to the nice colors (like running
# extraction again): '30 15;89 30;51 17;18 9;54 27;140 70;100 50;30
# 15;180 90;169 85;14 7;234 117;20 10;4 2; 150 75;12 6;2 1;160 80;10
# 5'

minclookup -discrete -lut_string '1 5;3 7;9 2'
Adelson_stx_labels.mnc.gz Adelson_stx_labels_extract.mnc.gz

/data/nil/david/public_bin/print_all_labels
Adelson_stx_labels_extract.mnc

# will calculate the volumes of all labels in that label file.

# To change cortex.obj files into PC friendly format: (Binary to
# Ascii)


/home/bic/david/public_bin/ascii_binary input.obj output.obj ascii