Eidors-logo    

EIDORS: Electrical Impedance Tomography and Diffuse Optical Tomography Reconstruction Software

EIDORS (mirror)
Main
Documentation
Tutorials
− Image Reconst
− Data Structures
− Applications
− FEM Modelling
− GREIT
− Old tutorials
Workshop
Download
Contrib Data
GREIT
Browse Docs
Browse SVN

News
Mailing list
(archive)
FAQ
Developer
                       

 

Hosted by
SourceForge.net Logo

 

GREIT Test Parameters

During the GREIT development process, we developed a set of figures of merit to characterize the performance of an algorithm.

Figure: GREIT test parameters: Amplitude (AR), Position Error (PE), Resolution (RES), Shape Deformation (SD), Ringing (RNG)
EIDORS code to run the GREIT evaluation is shown

Simulate 3D object

% Simulate obj $Id: GREIT_test_params01.m 2167 2010-04-04 21:39:48Z aadler $

fmdl = ng_mk_cyl_models([2,1,0.08],[16,1],[0.05]); 
fmdl.stimulation = mk_stim_patterns(16,1,[0,1],[0,1],{},1);
imgs= mk_image( fmdl, 1);

show_fem(imgs);
print_convert('GREIT_test_params01a.png','-density 60');

view(0,0)
xlim([-.4,.4])
zlim(1+[-.4,.4])
print_convert('GREIT_test_params01b.png','-density 60');


Figure: Simulation Mesh (right) and close-up of the electrodes (left)

Simulate Sequence of Targets

% Simulate obj $Id: GREIT_test_params02.m 4823 2015-03-29 15:17:16Z bgrychtol-ipa $

% Specify positions to simulate (only x-axis here)
r =  linspace(0,0.9,100);
xyzr = [r; zeros(1,100); ones(1,100);
     0.05*ones(1,100)];

[vh,vi] = simulate_movement(imgs, xyzr);

% Show GREIT images
opt.noise_figure = 0.5;
i_gr = mk_GREIT_model(fmdl,0.2,[],opt);
imgr = inv_solve(i_gr, vh, vi(:,1:5:100));
imgr.show_slices.img_cols = 5;
show_slices(imgr);

print_convert('GREIT_test_params02a.png','-density 60');


Figure: Illustration of the object locations in the simulation

Calculate the parameters

Image Reconstruction
% GREIT params eval $Id: GREIT_test_params03.m 2240 2010-07-04 14:41:32Z aadler $

% Reconstruct GREIT Images
imdl_gr = mk_common_gridmdl('GREITc1');
imgc{1} = inv_solve(imdl_gr, vh, vi);

% Reconstruct backprojection Images
imdl_bp = mk_common_gridmdl('backproj');
imgc{2} = inv_solve(imdl_bp, vh, vi);

fname_base = 'GREIT_test_params04';


Calculate performance
% GREIT params eval $Id: GREIT_test_params04.m 2240 2010-07-04 14:41:32Z aadler $

for i= 1:length(imgc);
   imgr = imgc{i};
   imgr.calc_colours.npoints = 32;
   params = eval_GREIT_fig_merit(imgr, xyzr);

   fname = sprintf('%s%c', fname_base, i+'a'-1);

   plot(r, params(1,:)); axis([0,0.9,0,2.0]);  ylabel('AR');
   print_convert([fname,'_ar.png'],  '-density 60',0.3);

   plot(r, params(2,:)); axis([0,0.9,-0.15,0.15]);  ylabel('PE');
   print_convert([fname,'_pe.png'],  '-density 60',0.3);

   plot(r, params(3,:)); axis([0,0.9,0,0.4]);  ylabel('RES');
   print_convert([fname,'_res.png'],  '-density 60',0.3);

   plot(r, params(4,:)); axis([0,0.9,0,0.3]);  ylabel('SD');
   print_convert([fname,'_sd.png'],  '-density 60',0.3);

   plot(r, params(5,:)); axis([0,0.9,0,0.6]);  ylabel('RNG');
   print_convert([fname,'_rng.png'],  '-density 60',0.3);
end







Figure: Evaluation of GREIT (right), and Sheffield Backprojection (left) GREIT parameters AR, PE, RES, SD, and RNG as a function of point radial position

Comment

There are several issues with this measurement, that should be kept in mind. This is an example of the use of the function, not a specification.
  • The object is just in one x position in the plane. It should be rotated to all x,y positions.
  • The various parameters are defined for a set of test positions for a given distance above and below the electrode plane, yet here we just test on the electrode plane.

Simulate Sequence of Targets above the plane (×25%)

% Simulate obj $Id: GREIT_test_params05.m 2167 2010-04-04 21:39:48Z aadler $

% Simulate positions that are 0.25 above the plane
r =  linspace(0,0.9,100);
xyzr = [r;
        zeros(1,100);
     1.25*ones(1,100);
     0.05*ones(1,100)];

[vh,vi] = simulate_movement(imgs, xyzr);

% Show GREIT images
i_gr = mk_common_gridmdl('GREITc1');
imgr = inv_solve(i_gr, vh, vi(:,1:5:100));
imgr.show_slices.img_cols = 5;
show_slices(imgr);

print_convert('GREIT_test_params05a.png','-density 60');


Figure: Illustration of the object locations in the simulation

Calculate the parameters

% GREIT params eval $Id: GREIT_test_params06.m 2240 2010-07-04 14:41:32Z aadler $

% Reconstruct GREIT Images
imdl_gr = mk_common_gridmdl('GREITc1');
imgc{1} = inv_solve(imdl_gr, vh, vi);

% Reconstruct backprojection Images
imdl_bp = mk_common_gridmdl('backproj');
imgc{2} = inv_solve(imdl_bp, vh, vi);

fname_base = 'GREIT_test_params06';

% GREIT params eval $Id: GREIT_test_params04.m 2240 2010-07-04 14:41:32Z aadler $

for i= 1:length(imgc);
   imgr = imgc{i};
   imgr.calc_colours.npoints = 32;
   params = eval_GREIT_fig_merit(imgr, xyzr);

   fname = sprintf('%s%c', fname_base, i+'a'-1);

   plot(r, params(1,:)); axis([0,0.9,0,2.0]);  ylabel('AR');
   print_convert([fname,'_ar.png'],  '-density 60',0.3);

   plot(r, params(2,:)); axis([0,0.9,-0.15,0.15]);  ylabel('PE');
   print_convert([fname,'_pe.png'],  '-density 60',0.3);

   plot(r, params(3,:)); axis([0,0.9,0,0.4]);  ylabel('RES');
   print_convert([fname,'_res.png'],  '-density 60',0.3);

   plot(r, params(4,:)); axis([0,0.9,0,0.3]);  ylabel('SD');
   print_convert([fname,'_sd.png'],  '-density 60',0.3);

   plot(r, params(5,:)); axis([0,0.9,0,0.6]);  ylabel('RNG');
   print_convert([fname,'_rng.png'],  '-density 60',0.3);
end







Figure: Evaluation for off plane object (25%) of GREIT (left), and Sheffield Backprojection (right) GREIT parameters AR, PE, RES, SD, and RNG as a function of point radial position

Last Modified: $Date: 2017-02-28 13:12:08 -0500 (Tue, 28 Feb 2017) $ by $Author: aadler $