johntmorgan / postmortem_mri_converter

This is a collection of several scripts that handle scanned .nii images from postmortem brains (including single hemispheres), converting them into full brains with coloration equivalent to a T1-weighted structural image from a living brain for further processing and analysis. We have successfully tested the outputs from this program on a recent version of Freesurfer.

This is a collection of several scripts that handle scanned .nii images from postmortem brains (including single hemispheres), converting them into full brains with coloration equivalent to a T1-weighted structural image from a living brain for further processing and analysis. We have successfully tested the outputs from this program on a recent version of Freesurfer.

The main program, PostmortemMRIConverter.py, has no external library dependencies, allowing it to be run under the PyPy interpreter, which produces a >100x performance increase, decreasing run times from several hours to a few minutes. NiBabel (and by extension Numpy) library dependencies are offloaded into NiftiLoad.py and NiftiSave.py.

As a tradeoff, you must run NiftiLoad.py before using this script to generate temp files that the script will operate on, and NiftiSave.py after running this script to convert back into .nii.

Note that temp files (-tmp) are generated but not deleted during this process. This allows the user to test variable settings without needing to reload each time. However, you will need to delete the -tmp files manually when you are satisfied with the output.

############################################################################# Setting up your computer to run this program

This program is written in Python 2.7. Please ensure that all installed libraries are up to date but not using Python 3+!

Step 1: Install Python 2.7.3 from here: http://www.python.org/download/releases/2.7.3/ On the page, select the Windows MSI installer (x86 if you have 32-bit Windows installed, x86-64 if you have 64-bit Windows installed.) I suggest using the default option, which will install Python to c:/Python27

Step 2: Install PyPy from here: http://pypy.org/download.html

Step 3: Install NumPy to your python directory from here: http://sourceforge.net/projects/numpy/files/NumPy/

Step 4: Install NiBabel to your python directory from here: http://nipy.org/nibabel/

Step 5: Copy the programs in this folder into the c:/Python27 directory You can also put them into another directory that is added to the Python & PyPy PATHs.

############################################################################# Steps to process a scan

Pre-run: If possible, load the image and strip it using a program like FSL. This is likely to greatly improve results. If the brain was embedded in gelatin, this program will handle it very poorly unless the gelatin has already been removed somehow!

Step 1: Load the .nii file and convert it into a format that PyPy can handle using NiftiLoad.py. From the command line, enter the directory where the programs are stored and type "python NiftiLoad.py" and hit enter.

Step 2 (optional): Run NiftiMirror if only a single hemisphere was scanned. You will probably need to run NiftiMirror and then save and check the file several times until you find settings that produce a good mirror. From the command line, enter the directory where the programs are stored and type "python NiftiMirror.py" and hit enter.

Step 3: Set variables below.

Step 4: Run this program from the command line by entering the directory where this file is saved and typing "pypy PostmortemMRIConvert.py"

Step 5: Save the .nii file to a new filename using NiftiSave.py. From the command line, enter the directory where the programs are stored and type "python NiftiSave.py" and hit enter.