bug fix for R2*

utils/ARLO.m

@@ -0,0 +1,54 @@
+%% function u = ARLO(y,x)
+%
+% Description: Fast monoexponential fitting by auto-regression
+% Samples have to be evenly sampled (i.e. even spacing)
+% At least 3 samples are needed
+% Assuming time series in the last dimension
+% ref: Pei et al. MRM 73:843-850(2015)
+% e.g. for function y=exp(-x/u), u can be estimated by ARLO
+%
+% Kwok-shing Chan @ DCCN
+% [email protected]
+% Date created: 8 October, 2016
+% Date last modified:
+%
+function u = ARLO(y,x)
+% ensure y is real
+% y=abs(y);
+
+% get dimension of y
+ndim = ndims(y);
+if ndim==2 && size(y,2)==1
+    y = permute(y,[2 1]);
+end
+matrixSize = size(y);
+N = matrixSize(end);
+
+% reshape y s.t. new y = [all y, time]
+y = reshape(y,[numel(y)/matrixSize(end)],N);
+
+% get the spacing
+dx = x(2)-x(1);
+
+% get sum of signal for i and delta i
+Si = zeros(size(y,1),N-2);
+deltai = zeros(size(y,1),N-2);
+for k=1:N-2
+    [Si(:,k), deltai(:,k)]= Simpson(y(:,k:k+2),dx);
+end
+
+% analytical solution for minimiser to obatain u
+a = sum(Si.^2,2);
+b = sum(Si.*deltai,2);
+u = (a + (dx/3)*b)./((dx/3)*a + b);
+
+% reshape u based on input dimension
+u = reshape(u,[matrixSize(1:end-1)]);
+
+end
+
+% Quadratic approximation of Simpson rule's in 4th order accuracy when J=2
+function [Si, deltai] = Simpson(y,dx)
+Si = (dx/3) * (y(:,1) + 4*y(:,2) + y(:,3));
+deltai = y(:,1)-y(:,3);
+end

utils/computeFiter.m

@@ -0,0 +1,41 @@
+%% function fiter = computeFiter(s,shat,NUM_MAGN)
+% s     - measured signal
+% shat  - simulated signal
+% NUM_GAGN - no. of phase corrupted echoes:
+% NUM_MAGN=0 : complex fitting
+% NUM_MAGN=length(s) : magnitude fitting
+% NUM_MAGN (0,length(s)) : mixed fitting
+%
+% Description: Compute the fitter for lsqnonlin
+%
+% Kwok-shing Chan @ DCCN
+% [email protected]
+% Date created: 
+% Date last modified:
+%
+function fiter = computeFiter(s,shat,NUM_MAGN)
+if NUM_MAGN == length(s) % Magnitude fitting
+    shat1 = abs(shat);
+    s1 = abs(s);
+    fiter = shat1(:) - s1(:);
+elseif NUM_MAGN == 0 % Complex fitting
+        fiter2 = shat(:) - s(:);
+        fiter2 = [real(fiter2); imag(fiter2)];
+%         fiter2 = [real(fiter2), imag(fiter2)];
+        fiter = fiter2;
+%         fiter = abs(fiter2);
+else
+    % Compute mixed fitting fit error
+    shat1 = abs(shat(1:NUM_MAGN));
+    s1 = abs(s(1:NUM_MAGN));
+    shat2 = shat(NUM_MAGN+1:end);
+    s2 = s(NUM_MAGN+1:end);
+
+    fiter1 = shat1(:) - s1(:);
+    fiter2 = shat2(:) - s2(:);
+    fiter2 = [real(fiter2); imag(fiter2)];
+
+    fiter = [fiter1;fiter2];
+end
+fiter = double(fiter);
+end