c4ir / fair.m Goto Github PK

On page 60, the book states that the transformation used from figure 5.2 (e) to figure 5.2 (f), a perfectly determine quadratic landmark-based map, "is invertible and thus meaningless for image registration purposes". This should say uninvertible instead of invertible, right?

Pull #2 breaks any example code that calls GaussNewton.m. For instance, E6_HNSP_MLPIR,SSD_affine2D.m breaks at line 65 in GaussNewton.m. There are actually two issues: (1) line 65 of GaussNewton.m has a syntax error and (2) the variable solver is being parsed as the string 'backslash'. Line 65 should be changed from error(1) to error('Error') because error() only accepts strings. A workaround for the second issue is to comment out lines 60-66 in GaussNewton.m. A fix will require further investigation into the parsing of the solver variable.

Hi,

I want to do rigid registration for 3D MRI and 3D CT of pelvis for the same patient using "Normalized Gradient Field". From the ebook on FAIR the links to tutorials don't work. Should I choose parts of the codes for sampling, interpolation etc from the ebook to run for my images or are these already implemented ?

I tried examples in Matlab, first I ran FAIRstartup.m and then E7_basic.m or E9_PETCT_MLIR_NGF_mbCurv.m but I get the same error as following:

Unrecognized function or variable 'getCellCenteredGrid'.

Error in jpgs2data>@(k)getCellCenteredGrid(omegaI(k),m) (line 64)
xc = @(k) getCellCenteredGrid(omegaI(k),m);

Error in jpgs2data>@(I,k)viewImage(imgModel(I,omegaI(k),xc(k)),omegaI(k),m) (line 66)
viewData = @(I,k) viewImage(imgModel(I,omegaI(k),xc(k)),omegaI(k),m);

Error in jpgs2data (line 69)
subplot(1,2,1); viewData(dataT,1); title('template');

Error in setup2DPETCTData (line 25)
expfile = jpgs2data('','PET-CT-PET.jpg','PET-CT-CT.jpg', ...

can someone help me please!

Thank you for such a great work!

I'm currently trying to register a projective distorted picture, but I couldn't find the very transformation.

May I ask for projective2D.m for Parametric Image Registration?

I recommend following changes :

In the function routine mfPu, the filter function at line number 163 should be
C(:,1,1) = [1 2 1]/4
The division should be by 4 instead of 2 so that the total weight is 1.

On the similar note, the function in TriMesh3 (inside add-on FAIRFEM) should be (from line number 346-367) :

% include parent nodes
Pyn = 0.5*yn(1:2:end,1:2:end,:);
% distribute in x direction
t = yn(2:2:end-1,1:2:end,:);
Pyn(1:end-1,:,:) = Pyn(1:end-1,:,:) + (1/16)*t;
Pyn(2:end ,:,:) = Pyn(2:end ,:,:) + (1/16)*t;

% distribute in y direction
t = yn(1:2:end,2:2:end-1,:);
Pyn(:,1:end-1,:) = Pyn(:,1:end-1,:) + (1/16)*t;
Pyn(:,2:end ,:) = Pyn(:,2:end ,:) + (1/16)*t;

% distribute cell-centers
t1 = yc(1:2:end-1,1:2:end-1,:);
t2 = yc(2:2:end ,1:2:end-1,:);
t3 = yc(1:2:end-1,2:2:end ,:);
t4 = yc(2:2:end ,2:2:end ,:);
Pyc = 0.5yn(2:2:end,2:2:end,:) + (1/8)(t1+t2+t3+t4);
Pyn(1:end-1,1:end-1,:) = Pyn(1:end-1,1:end-1,:) + (1/16)*t1;
Pyn(2:end ,1:end-1,:) = Pyn(2:end ,1:end-1,:) + (1/16)*t2;
Pyn(1:end-1,2:end ,:) = Pyn(1:end-1,2:end ,:) + (1/16)*t3;
Pyn(2:end ,2:end ,:) = Pyn(2:end ,2:end ,:) + (1/16)*t4;

Please also find a test script and data set attached.
data.zip

Hi, it is a wonderful toolbox. I have a problem. It told me 'the function E10_2Ddisc2C_hyperElastic can not be recongnized' when running the file GaussNewtonLDDMM. In fact, I can not find the function E10_2Ddisc2C_hyperElastic.m in the FAIR toolbox.

An older version of the repository provided the file getCellCenteredGrid.m in the folder kernel/numerics, I can't find it in the newest version.