11.3 MB



This shows the MRI within a manually generated transform. The goal here is to achieve a rough initial alignment such that the key structures of interest overlap enough for running an automated registration.[br]Note that there are very few structures common to both images, namely kidney, vertebrae and some vasculature. We use these as landmarks for the manual alignment and also to define the masks in the next step.[br] ;2. Manual Initial Alignment ;This shows the original CT and MRI in their original unregistered orientation. Note that the field of view (FOV) and scan orientation are very different for the two volumes. The MR is rotated by more than 90 degrees and the CT's field of view is significantly smaller than that of the MRI. Automated registration will require both initial alignment and masking to succeed.[br]We will register the MR to the intra-op CT, the common rationale being that the pre-op MRI contains better contrast and visibility of the pathology (tumor) than the CT and thus can help in guiding the intervention. ;1. unregistered ;Scene at MRML file save point ;Master Scene View ;Scene at MRML file save point ;Slicer Data Bundle Scene View ;This shows the result of the automated affine registration. We used the previously obtain rigid transform as starting point, and the same masks. Note that both kidney and vertebrae are now reasonably well aligned.[br] ;5. Affine: final registration ;This shows the result of a rigid automated pilot registration. The purpose here is to see if the image content provided is sufficient for the algorithm such that it converges to a stable result. If the result of this rigid step is close but different from the provided intial manual starting point, we can proceed to higher DOF. If not, we need to revisit the masks and other registration parameters.[br] ;4. Rigid automated registration: testing stability ;This shows the manually generated masks for the CT and MRI. Goal is to restrict the focus of the automated registration to areas we know contain common structures. Since the field of view of the CT is limited, we focus on the kidney, the main vertebrae and the vena cava as the main content. The masks need not be overly precise. We use a large (~50mm) brush tool to generate the masks for both volumes. ;3. Masks ;





Created 08/20/2013
Uploaded by Jean-Christophe Fillion-Robin
Views 149
Downloads 26
Revisions 2
Files 1


Rating Distribution