Fork me on GitHub

Registration Workspace


Overview, what can be done?

The Registration Configuration is the interface through which all the registraton plugins can be used. Our video tutorial page also illustrates some of the interface elements documented here. As a reminder, for readers not used to the concept of registration, let's introduce shortly the concept.

When an image is acquired, it is placed in a given coordinate system where each voxel is aligned on a grid. Any other acquired image will be placed in a similar grid, but the subjects may not be placed at the same position, or with the same orientation in this coordinate system. Different transformation can happen during acquisition between images. The same subject taken twice won't be aligned with the first one.


All registraton algorithms will have the goal of realigning a moving image on a reference or fixed image, and output a newly interpolated image, realigned on the same grid as the fixed image.

The concept can be extended to meshes, time sequences, or a mix of images and meshes. medInria 2.x will try to address all the use cases. At the moment of the realease of medInria 2.0, only image to image (3d volume) registration was available, but more algorithms will follow. Time sequences, in which each frame is aligned with the first one, is actively being worked on.


The configuration in itself receives, on the right hand side, the toolboxes to enter the parameters of each available algorithm. The registration algorithms being very different to one an other (affine, non rigid deformation field...), the toolboxes in question can host very different parameters indeed. In the middle, there are 2 sets of containers, the Compare layout, and the Fuse layout. They allow the user to set the inputs and watch the results of a registration process.

 registration configuration


Compare and Fuse layouts

Compare Layout

This container layout is the first thing the user sees in the registration configuration. It is composed of two view containers. One is labelled "fixed", the other "moving". The user will drag from the navigator and drop into the corresponding containers the two needed volumes.

To check the alignment of these volumes, before as well as after  processing the moving image, the user can link the two view by clicking on the chain-link (icon B in the next figure) icon of the views (more information in the visualization documentation). To ease the comparison, the views can display the axis by clicking on the axis check box in the view properties toolbox.

chainlink                                    axis Icon

The result of the registration process will take the place of the moving image after computation. This way a first rigid registration process can be followed by a second non rigid step for instance.


Fuse Layout

Some will prefer to overlay the images to compare them before and after registration. The fuse layout is there for this reason. It automatically shows the two inputs overlayed within the same view. The user can set the transparency of each layer with the help of a slider, or switch from one volume only or the other in one click. It can also give different transfert functions (Look up tables) to each volume. For more information on presets check the visualization documentation.

2 colours Before Registration


2 colours After Registration 


How to run a registration algorithm

  1.  Choose your algorithm in the list in the registration toolbox. Another toolbox with the specific parameters will show up.
  2. Set up all the needed parameters
  3. Drag and drop your two volumes if not already done.
  4. Click on run.
  5. After a while the processed volume shall replace the moving one.

Once done you may check the accuracy of the registration using either the compare layout and the axis, or the fuse layout, switching between layers or changing their colours.

Please note that you can get information on the plugin by click the "i" button in the toolbox's header. You should get links to the authors and to more detailed documentation about the algorithm itself.


Back to Documentation Index