International Journal of Biomedical Imaging / 2009 / Article / Tab 2

Research Article

Reconstruction for Time-Domain In Vivo EPR 3D Multigradient Oximetric Imaging—A Parallel Processing Perspective

Table 2

Pseudocode showing the reconstruction and oximetry thresholding (stages 2 and 3 of the parallel system).

( )Set parallel oxygen thresholding on the GUI
( )A filtered data set is chosen from the shared directory of the server
( )Start time point, increment step value, total time points, dead time, zero-factor are provided
( )Raw data of 12 time points (4 time points per gradient) are selected from the filtered gradient data by calling a Mex C code and stored as three binary TPS files in the same shared directory
( )For every gradient, the raw data of the 4 time points are read from the binary files into a variable and perform the steps 6 and 7
( )Baseline correction is performed on the each of the 4 time point data
( )FOV is computed and zero-filling is performed
( )3D FFT-based reconstruction is performed on the time point data to generate twelve 3D spin density images
( )3D mesh view of the sixth spin density image is displayed
( )A spin threshold value for oxygen computation is input on the GUI
( )Create and configure local scheduler and create a parallel job for line width and oxygen computation
( )Assign number of workers to 4 and assign the oximetric threshold code to the workers
( )Create task objects for the parallel job and run the parallel job to job queue
( )Store the LW images from the results of the parallel job, in matrices
( )Oxygen images are then computed from the LW images
( )The sagittal, axial, and coronal planes of both spin density and oxygen images are displayed on the GUI simultaneously
( )The steps from 11 to 17 can be repeated for a different spin threshold value