Eliminating for-loops (compact notation?)
2 views (last 30 days)
Show older comments
I am trying to compare fMRI images, but I seem to have quite an inefficient comparison algorithm.I have heard that efficiency gains can be made by using the slightly more native compact notation, but I have no idea where to start with this. Here is the code, it will end up being run several million times.
for xInd=1:xMax
for yInd=1:yMax
for zInd=1:zMax
if ~isnan(imageArray_1(1,xInd,yInd,zInd))
statImage(xInd,yInd,zInd)=abs(log(mean(imageArray_1(:,xInd,yInd,zInd)./imageNormalisationDivisor_1))-log(mean(imageArray_2(:,xInd,yInd,zInd)./imageNormalisationDivisor_2)));
end
end
end
end
Essentially there are 10 three-dimensional arrays per condition, and the difference in log mean value at each voxel across each condition is compared. Each image is normalised by dividing it by the mean value.
It's not a dealbreaker if it can't get much more efficient, I can leave it running for the day or so it would need, but I have a feeling that there are improvements that could be made.
Thanks, Matt
0 Comments
Accepted Answer
Thorsten
on 28 May 2015
Edited: Thorsten
on 28 May 2015
Oh, that's easy (just kidding):
X = squeeze(abs(log(mean(imageArray_1, 1)./mean(imageArray_2,1)*imageNormalisationDivisor_2/imageNormalisationDivisor_1)));
Note that the length of this one-liner is partly due to your long variable names. Compare
X = squeeze(abs(log(mean(I1, 1)./mean(I2,1)*s2/s1)));
To compare both solutions, use
max(abs(statImage(:) -X(:)))
That's not zero, but should be a small value in the range of eps, the machine precision.
1 Comment
More Answers (0)
See Also
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!