How can I change a pixel value of a CT image series to display HU.

How can I change a pixel value of a CT image series stored as a variable in Matlab format to HU units,.... as in how can I subtract 1024 from all the pixels so they display in matlab as HUs?

Answer 1

[Nick]

Your question is not really clear to what your intent is. If you just want to subtract 1024 of each pixel it would be:

HUImage = CTImage-1024;

Ofcourse you would need to make sure you have a value type that can be less that 0 so you could make sure it is capable of being signed for example int16's or doubles:

HUImage = int16(CTImage)-1024; % this is assuming your CT image is something like: uint16 or so which seems unlikely
% or
HUImage = double(CTImage)-1024;

% then display it using something like:
imshow(HUImage, []) % rescale to min and max value
% or
imshow(HUImage,[-1000 2000]) % set display range yourself, outliers will be black for lower than -1000 or white for higher than 2000

But houndsfield units are not just subtracting a certain value from some image, it is a relationship of the average linear attentuation coefficient inside a voxel compared to this of water and air:

If your CTImage is a doubles array of linear attenuation coefficients then you need to know the linear attentuation coefficients of water and air and could turn the image into HU using:

uWater % linear att water
uAir % linear att air -> uAir is so much smaller than uWater that u can prolly ignore this
linearAttCoef = double(CTImage); % already assuming its double but just putting it here
% calculate HU from linear attenuation coefficients
HUImage = ((linearAttCoef - uWater)./(uWater - uAir)) .* 1000; % HU unit conversion formula

Hopefully this clears out enough for you, without more information on your problem we cannot answer your question fully.

[Dan Smith]

Thank you for your reply. I ran CERR on CT dicom files to create one mat file that has a variable planC containing the image data. I then took planC (a 1x25 cell) with code imagesc(CTimgmat(:,:,200)) to display the image and compared the x,y,index position/value to that which is displayed for the dicom images in OSIRIX or any other viewer that displays the pixels in HU units. The two didn't match up by about a thousand but it differs pixel to pixel. My goal is to run imagesc(CTimgmat(:,:,"slice#")) and have index # display as HU.

I applied your equation above but values still don't match up to HU