Dear Feng,
the tissue volumes are only obtained inside a predefined mask to obtain a more reliable value. This is achieved by applying some morphological operations and slightly smoothing the mask:
Yp0 = single(P(:,:,:,3))/255/3 + single(P(:,:,:,1))/255*2/3 + single(P(:,:,:,2))/255;;
Yp0(smooth3(cat_vol_morph(Yp0>0.3,'lo'))<0.5)=0; % not 1/6 because some ADNI scans have large "CSF" areas in the background
Yp0 = Yp0 .* cat_vol_morph(Yp0 & (Ysrc>WMth*0.05),'lc',2);
Yp0toC = @(Yp0,c) 1-min(1,abs(Yp0-c));
cat_err_res.init.subjectmeasures.vol_abs_CGW = [prod(vx_vol)/1000 .* sum(Yp0toC(Yp0(:),1)), ... CSF
prod(vx_vol)/1000 .* sum(Yp0toC(Yp0(:),2)), ... GM
prod(vx_vol)/1000 .* sum(Yp0toC(Yp0(:),3)), ... WM
prod(vx_vol)/1000 .* sum(Yp0toC(Yp0(:),4))]; % WMH
Best,
Christian
>Dear CAT experts,
>
>I have a question about the TIV calculation in CAT12 software.
>
>I want to calculate the GM/WM/CSF manually and examine whether the values are equal to that obtained from the software. Basically, the stuctural MR image was segmented using VBM and then p1*(gm),p2*(wm) and p3*(csf) files were obtained. Then, all the values in p1/p2/p3 image were added, multiplied the voxel volume (product of voxel size x,y,z) and divided 1000. Finally, the GM/WM/CSF volume (unit: ml) was obtained, and these values are exactly equal to that in the pxxx_seg.txt.
>
>However, when I use the same aforementioned calculation procedure in CAT12, the GM/WM/CSF volume value I obtained are not equal to that obtained from CAT12 (catreport*.pdf). I am wondering the difference in TIV (GM+WM+CSF) calculation between VBM8 and CAT12, and how could I calculated the GM/WM/CSF manually?
>
>Best,
>Feng
|