<div dir="ltr"><div><div>Hi Philip,<br></div>thanks for quick answering,<br><br></div><div>yes I followed that convention.<br><br>It is rather complicated visualizing the issue in any way, because, for instance, pdview looks good. <br>However, i made a test right now with a DTI dataset of mine. 32 directions and a b0 image. <br>This are commands I used<br><br></div><div>modelfit -inputfile dwi.Bfloat -model nldt_pos .schemefile scheme.scheme -bgmask Mask.Bshort > DTI.Bdouble<br></div><div>cat DTI.Bdouble | fa | voxel2image -outputroot FA -header DTI.nii<br></div><div>cat DTI.Bdouble | dteig > EIG.Bdouble<br></div><div>voxel2image -inputfile EIG.Bdouble -components 12 .outputroot eigs_ -header DTI.nii<br><br></div><div>then, I used spm tool to import nifti images in matlab, and counted how many rotation matrices had determinant equal 1 or -1 (within the foreground); I found 55.21% of voxel having proper rotation while 44.79% having determinant equal -1.<br></div><div><br>If you like I can send you eigensystem, exitcode volume, mask and code I used on matlab so that you might check if I did something wrong.<br></div><div><br>Attached a representative axial view of my dataset, showing that fitting was fine..<br></div><div><br></div><div><br></div><div class="gmail_extra">Alessandro<br><div dir="ltr"><div><div dir="ltr"><div><div dir="ltr"><div><br></div></div></div></div></div></div>
</div></div>