EIDORS: Electrical Impedance Tomography and Diffuse Optical Tomography Reconstruction Software

Hosted by

GREIT algorithm candidates

function [img,map]= ALG_NAME( ref_meas, reconst_meas )

img	32×32×N conductivity change image representing the N measurement instants
map	32×32 vector indicating which pixels are inside the medium (=1) and which are outside (=0).
ref_meas	208×1 measurement vector representing the background conductivity
img	208×N measurement vector representing the N measurement instants

Difference Imaging

A difference imaging algorithm may look like this example:

function img= ALG_NAME( ref_meas, reconst_meas )

   load ReconstMatrix_from_ALG_NAME RM;

   % Expand ref_meas to the full size of reconst_meas
   num_meas = size(reconst_meas,2);
   ref_meas = ref_meas * ones(1,num_meas);
   dv = reconst_meas - ref_meas;

   % reconst image
   ds = RM*dv;

   img= reshape(ds, 32,32,num_meas);

Normalized Difference Imaging

A normalized difference imaging algorithm may look like this example:

function img= ALG_NAME( ref_meas, reconst_meas )

   load ReconstMatrix_from_ALG_NAME RM;

   % Expand ref_meas to the full size of reconst_meas
   num_meas = size(reconst_meas,2);
   ref_meas = ref_meas * ones(1,num_meas);
   dv = ( reconst_meas - ref_meas ) ./ ref_meas; % CHANGE IS HERE:

   % reconst image
   ds = RM*dv;

   img= reshape(ds, 32,32,num_meas);

Comments

• An algorithm accepts no parameters. There is no place to enter a regularization hyperparameter, for example. This is intentional, since each GREIT algorithm is supposed to be completely stand alone. Different variants of algorithms should have different names.
• Algorithms do not accept shape information. Again, this is intentional, since GREIT algorithms for different shapes (adult thorax, cylindrical tank, etc.) are considered to be different algorithms.

Algorithm examples

• Sheffield Backprojection
• NOSER-type difference and normalized difference