Living Image 3.1

Living Image ® Software User’s Manual
Table B6 3D analysis preferences
Item Description
N Surface Limits The maximum number of surface intensity points to use in the
reconstruction at a given wavelength. The range is 200 to 800 and the
default is 200. The time required for reconstruction is shortest for smaller
values of N (for example, 200). However, a large N value may give a more
accurate result because more data are included in the fit.
Voxel Size Limits Voxels are the small cubes of space inside a subject, each of which
contains a light source (much like a pixel in a 2D image). The DLIT
reconstruction begins with large voxels, specified by the voxel size limit
(the length of a side of the voxel cube in mm). At each iteration, the
algorithm reduces the size of the voxel by a factor of two until the
optimum solution is found.
The voxel size limits are a minimum of five and a maximum of 10. The
default range is set to 6-9 mm. A larger range of voxel limits ensures a
more reliable solution, but requires more computational time. The default
range of 6-9 is usually adequate to determine the optimum solution.
Voxel Size
Increment
Uniform Surface
Sampling
NNLS + Simplex
Optimization
This is the step increment in voxel size, stepping from the minimum voxel
size limit to the maximum voxel size limit. For example, if the voxel size
limit ranges from 6-9 mm, a voxel size increment = 1 gives four starting
voxel sizes (6, 7, 8, and 9 mm).
The default increment of 1 mm is usually adequate, however smaller
increments can be used if you want to sample finer voxel sizes. Smaller
increments will significantly increase the time required for
reconstruction.
If this option is chosen, the surface data for each wavelength will be
sampled spatially uniformly on the signal area. If this option is not chosen,
the maximum ‘N surface elements’ will be sampled for the data. This
means that the N brightest surface elements will be used as data in the
reconstruction. Typically, non-uniform sampling is recommended if there
is a single bright source, while uniform sampling is preferred if there are
several scattered sources.
If this option is not chosen, the software uses a linear programming
algorithm to seek the solution (Simplex solution). If this option is chosen,
the software also applies a non-negative least squares optimization
algorithm at each iteration to provide a better solution for source power.
The Simplex solution is more robust, but tends to underestimate the
source flux in each voxel. Therefore, the NNLS + Simplex option is
recommended.
NNLS Weighted Fit Choose this option to weight the wavelength data proportionally to its
intensity in the NNLS reconstruction. This option is especially useful if the
intensity of longer wavelength data is orders of magnitude greater than
the intensity of shorter wavelength data.
189