| Description |
This function implements a reverse diffusion scheme to interpolate images to correct for partial volume effect. Under the assumption that the pdf is about the size of the pixel (e.g. MRI), edges between two structures can be recovered.
There is a script (go1.m) that asks to select multiple dicom files and then runs the main function. Any file formats could be read upon a small modification (imread instead of dicomread).
% PVcorr2D: partial volume correction with reverse diffusion
%
% [I,Flow,H] = PVcorr2D(Ir,ratio,option);
%
% Ir: image to be interpolated
% ratio: 2 for x2 interpolation
% option: structure with parameters
% option.Niter = 200; Max number of iteration
% option.tol = 0.001; tolerance for stopping
% option.f_H = 1; if 1, compute the entropy
% option.f_display = 1; if 1 dispaly intermediate results
% option.f_PVE = 1; if 1 use flow constraint
% option.ordmax = 5; neighbor to compute max flow (out of 8 neighbors)
% option.ordmin = 4; neighbor to compute min flow (out of 8 neighbors)
% for the two last options, one can try 6 and 3, or 7 and 2
%
% I: corrected image
% Flow: flow across iterations
% H: entropy across iterations if option.f_H is set to 1
%
% OS, Case Western Reserve University, 16jul04
%
% From the attached paper:
% Olivier Salvado, Claudia M. Hillenbrand, and David L. Wilson,
% ?Partial Volume Reduction by Interpolation with Reverse Diffusion,?
% International Journal of Biomedical Imaging,
% vol. 2006, Article ID 92092, 13 pages, 2006. doi:10.1155/IJBI/2006/92092
% |
