Rigid Registration Process

Figure 5.3: This screen shot shows the two views in full screen mode where the zoom interaction has been used to isolate a specific area. A manual rigid registration is being performed (the ``Control Points'' button has been pressed). Yellow crosses are the 3D spatial points that have to correspond between images. Yellow circles define the directions (following the blue lines). Note that no matter the circle is far from the cross, only the direction is taken in account.

This module provides easy-to-use manual rigid registration. The way to do so is simple : the user chooses a single 3D point and a direction for each image that should correspond one to the other. Then the RegistrationTool turns this information into a rigid matrix. The calculated rigid matrix is first applied to the moving image. Then the moving image is re-sampled to meet the fixed image size and spacing information.

Here is an easy to run process to perform rigid registration manually :

• 1. First select the view orientation that make appear the largest difference between images.
• 2. Deselect the link mode and navigate in the images so that the slices correspond for both images.
• 3. You may use the full screen mode and use the zoom interaction to specify the problematic area.
• 4. Press the green button on the control panel (Fig. 5.3).
• 5. Left click on the fixed image main view on the desired 3D point. Then the drawn line indicates the desired direction (no matter how far from the 3D point it is). Left click again when the desired direction is reached. (rotations of more than are not allowed : it changes the orientation of the image). Please refer to Fig. 5.4
• 6. Repeat 5. in the moving image main panel. If you need to do another interaction in the views (such as navigation or window-level change) you may press again the green button once again to deactivate the control point interaction.
• 7. As soon as the desired positions and directions are reached for both images, you can press ``Perform'' to perform registration. The positions and directions information is turned into a rigid matrix and applied to the moving image. If you want to go back to the first configuration, press ``Reset''. Careful you will loose any previous registration.

IMPORTANT : Once the moving image is registered, both images are automatically position-linked one to the other. To switch off this link mode, press the ``link/unlink'' button in the toolbar. Note that the moving image is also automatically re-sampled into the fixed image size and spacing information. The re-sampling process uses a linear interpolation (default) between pixel values. As you can see in the control panel (Fig. 5.4) you can choose another type of interpolation. Note that the B-spline and sinus cardinal interpolations takes longer to be done.