McStrip: Minneapolis Consensus Stripping - V1.1

McStrip processing stages: raw, coarse mask, thresh mask, BSE mask, McStrip mask

Corresponding author: 

What it does:

McStrip (Minneapolis Consensus Strip) is a hybrid method for removing skull and scalp from a T1-weighted MRI brain volume. The output is a binary mask that includes cortical grey matter, white matter, ventricular CSF, and some subarachnoid CSF.   Experiments have shown the hybrid mask to be a closer match to manual strips than the masks of any of its component methods (see references below).

• Operates on ANALYZE format volumes (8 and 16 bit integers).
• Computes final and four intermediate brain masks which are made by each method in the hybrid algorithm.
• A tuning program is provided to assist in adjusting operating parameters to suit your datasets.
• Runs on PseudoColor and TrueColor display systems.

• IDL5.2 or higher as either a development or runtime license.  The IDL6.0 Virtual Machine (it's free!) may also be used.  Obtain IDL from Research Systems Inc. at  http://www.rsinc.com/
• AIR5.0 or higher.  Obtain this from Dr. Roger Woods at http://bishopw.loni.ucla.edu/AIR5/
• BSE2 or BSE3.  This software is distributed as a part of BrainSuite  http://neuroimage.usc.edu/
• Requires 3.5 MBytes disk space for software, 55 MBytes for template files, and a working area of ~8 times the size of a subject volume.

• Version 1.0 is only provided as IDL SAVE sets; if you encounter errors, I want to hear from you!
• Version 2.0 will be released as IDL source code.
• Save sets are available for IDL versions 5.2, 5.3, 5.5, 5.6 and 6.0.  The 5.5 and 6.0 save sets have been verified under Windows 2000 as well as Linux.
• The IDL6.0 save set has been run using an IDL Virtual Machine on Windows 2000, Linux and SGI IRIX.

• The current version is 1.1 (December 16, 2003).  I corrected a problem for users running on SGI IRIX.
• The prior version was 1.01 (November 11, 2003).  I corrected a problem for users running /bin/csh instead of /bin/tcsh or /bin/bash.

• No testing has been performed on Macintoshes.
• Subject volumes are expected to be in ANALYZE format, where the first, second and third dimensions correspond to sagittal, coronal and axial slices respectively.
• Computation time using IDL5.2 will be considerably slower than the other versions due to lack of some built-in functions.
• Computation times for the example volume provided were 28 minutes on a 500MHz processor and 8 minutes on a 2GHz processor.

• Use this REQUEST FORM to request the software.
• The (16 bit) checksum for the tar file is ________.

Rehm K, Shattuck D, Leahy R, Schaper K, Rottenberg DA. Semi-automated stripping of T1 MRI volumes: I. Consensus of intensity- and edge-based methods.  NeuroImage. 9(6)S86, 1999.

Rehm K, Schaper K, Anderson J,  Woods R, Stoltzner S, Rottenberg D.  Putting our heads together: a consensus approach to brain/non-brain segmentation in T1-weighted MR volumes.  Submitted to NeuroImage, November, 2003.