Overview of PMI Toolbox

The Photon Migration Imaging (PMI) toolbox implements numerous algorithms for performing the forward and inverse calculations associated with diffusive optical imaging. Some of the features supported by the PMI Toolbox include:

• Infinite, semi-infinite, and infinite-slab geometries
• Homogeneous and inhomogeneous media
• Simulation routines for modeling DOT experiments
• Fitting data for background optical properties
• Time-domain and frequency-domain imaging
• Interfaces with Monte Carlo and finite difference forward solvers
• Linear and non-linear image reconstruction
• 2-D data visualization and 3-D volume visualization
• Data calibration

Using the PMI Toolbox

The PMI Toolbox runs under Matlab, which supplies many of the supporting routines used by the toolbox. While most routines will run without any toolboxes outside the core Matlab package, a small number of routines use functions from other Matlab toolboxes. In particular, some of the routines for fitting background optical properties use routines from the optimization toolbox (sold by Mathworks as a separate package).

To use the PMI Toolbox, the user first sets up a pair of structures, traditionally called SD and Medium. These structures can either be generated by hand or extracted from the data file (if the data is stored using the PMI data format). Having generated the SD and Medium structures, the PMI Toolbox provides standard routines for calculating the forward problem, inverting the data or (for simulations) generating the data, and then visualizing the results.

The SD and Medium Structures

All the information about the imager/medium system being modeled is contained in two Matlab structures: the SD and the Medium structure. The SD structure holds all the information about the imager (wavelengths, times, frequencies, probe geometry, etc.). A complete copy of the measurement list is also included as part of the SD structure (see below). The Medium structure, on the other hand, describes the medium being imaged. In here are kept the optical properties, the size of the volume to be reconstructed, and the experimental geometry.

The Measurement List

The measurement list is used to represent the experimental parameters corresponding to each measurement. Each row of the measurement list corresponds to a single measurement; the columns contain indicies into different tables in the SD structure. This format also makes it very easy to pass sub-sets of the measurement list to the other routines in the toolbox (for example, to save time you might choose to reconstruct images using only one laser wavelength). The ability to easily include or exclude measurements, based on any criteria the user desires, is one of the most important features of the PMI toolbox.

Working with Experimental Data

While Matlab's file I/O routines are powerful enough to make importing almost any type of data easy, the preferred data format for storing data is the native PMI data format. The PMI data format is a very general data format and can be used to represent any experimental geometry supported by the toolbox (and can be extended to include unsupported auxiliary information as well).

The advantage of the PMI data format over may other data formats is that, in addition to the experimental data, the PMI data format includes meta-data about the experimental geometry (source and detector locations, imager wavelengths, measurement order, etc) used to generate the data. Including meta-data reduces the opportunities for mistakes and confusion when re-analyzing data after a period of months or years.

A general set of routines for reading and writing PMI formatted data files as part of the toolbox.