A new release (v3)

I posted the last public jip version on the NITRC site almost 2 years ago (version 2.1). There have been some small but significant changes that have been incorporated into the newest version. The major changes in version 3 are listed here.

 

Display program:

  • On startup, the old display software didn’t work very well with very large volumes.  If the volume could not fit on the screen as a mosaic, it would try to use the tri-plane format, and then go to a single slice if the 3-plane format still was too large. Now, you get what you ask for.  If a large volume doesn’t fit onto your screen using a 1-to-1 pixel mapping, then it will be shown at lower resolution.
  •  A grayscale bar has been attached to the right side of the image window to facilitate windowing of images.

 

Alignment program

  • The old distortion fields used shifted cosine-squared spatial basis functions, which have an advantage of smooth variation. However, one cannot easily develop a diffeomorphic multi-scale algorithm with those bases, because distortion fields cannot be represented consistently at each spatial scale. The new algorithm uses a 1-2-1 weighting of nearest neighbors at the highest resolution for the basis functions, and then lower-resolution bases are constructed from higher-level bases using a mixing matrix. The upshot is that distortion fields are represented consistently at each spatial scale, and this prevents bad mappings from creeping into the final solution.
  • The target and source dimensions and resolutions now are stored in the output alignment file, so that one can use inverse transformations if desired.

 

PET SRTM using GLM

We have been performing simultaneous PET/fMRI studies here, so I have created a GLM version of the common simplified reference tissue model (SRTM) for analysis of PET data using tracers that are amenable to this approach.


I'm sure there are things I am forgetting.  If I remember, I will update the list.  Additionally, I have created a few other programs, such as aligning histological slices into volumes using MRI as a template.  Other programs may be this added to this set at a later time.

Joseph B. Mandeville, Athinoula A. Martinos Center for Biomedical Imaging at MGH/MIT/Harvard