Objects fusion enables to create a new object from 2 or more other objects. Indeed, if you only put two volumes in the same window, you will see only one. To see the two volumes, you need to mix voxels from the two volumes in order to obtain a new volume. Note that fusionning more than 2 objects is only possible since 1.30 version. Besides, several new features have been added for fusion management. Several fusion combinations between objects are available,but for the moment, let's see a fusion between two 3D volumes for example :
STEP 1 : Load the images to merge. Here, we will fusion an anatomy and the brain mask obtained from BrainVISA anatomical pipeline.
STEP 2 : Select the two volumes in objects list with Ctrl + bouton gauche.
STEP 3 : Then click on fusion button
.STEP 4 : A new window pop up to select objects order and fusion type (fusion types offered differs according to selected objects, this will be detailled later, in advanced part of the manual).
STEP 5 : Click on Ok to create the new Fusion2D object.
STEP 6 : Put the Fusion2D object in a window.
STEP 7 : If the Fusion2D object is all in black, you must change fusion mode. So right click on Fusion2D object to get its menu. Choose Fusion => Control 2D fusion. This window opens :
STEP 8 : You can change the mapping mode. The default is Linear : it does a linear combination of the two volumes. The Geometric mode does RGB channels multiplication. For linear fusions, you can set objects transparency with the cursor Mixing rate.
STEP 9 : By default, the 2 volumes will have the same palette. To change at least one, do Right clik menu on a volume => Color => Palette.
NOTE : Since the last version of Anatomist, it is possible to create multi-fusion, that is to say fusionning more than 2 objects. For volumes, here is the method : with for example 3 volumes (V1, V2 and V3), Anatomist actually fusion the last volume and the volume above in the list (V2 and V3 gives V23). Then, from this fusion object, it creates a new fusion with the volume above (V23 and V1 gives V23_1). To set parameters for each fusion, you have to select the fusion's second volume. For example, to set parameters for fusion V23_1, you must select V2.
NOTE : In this example, we didn't have to matter about objects referential management because the brain mask (brain_lesson1.ima) have been generated from the anatomy, so objects are in the same referential. But if we had done a fusion between an anatomical volume and an activation map (which is in another referential since this map comes from a functional volume), we would have to handle referentials to put the objects in a coherent coordinates system.
The following table show which type of fusion is available according to the type of objects :
Table 7.1. Fusion descriptions
| Objects | Fusion name | Description |
|---|---|---|
Only one image  | FusionSliceMethod | Fusion allowing to cut a volume across itself: to view/intersect 2 different slices of the same volume in the same window. |
2 or many images  | Fusion2DMethod | The volumes are merged in one volume. A voxel of the resulting volume is a combination of the same voxel in each original volume. |
Volume + Mesh  | Fusion3DMethod | Maps on the mesh a texture corrsponding to the volume values. |
Volume + Mesh | FusionCutMeshMethod | Mesh cut by a plane: the cutting plane will have the texture of the volume slice. When you put this object in a 3D window, the "cut mesh" control is available. It enables to control the orientation of the slice (shift) and its position (contrôle) against the mesh. |
2 textures | FusionTextureMethod | Creates a 2D texture from two 1D textures. |
2 textures | FusionMulitTextureMethod | Multi-texture: allows to map several textures on a mesh. |
Mesh + Mesh | SurfaceMatcher | Matching surfaces. This object gives access to a surface deformation algorithm. It tries to transform one surface into the other. |
Mesh + Texture | FusionTexSurfMethod | Textured surface. |
- Fusion mode : Geometrical, linear, rate.
- Methods of interpolation : the method to estimate the value for the intersection between the mesh and the volum at each point.
Table 7.2. Interpolation methods
Section Description Point to point the simplest: only the information coming from the voxel directly under the mesh vertex is used, directly. Do not use the depth and the step prameters.
Point to point with depth offset (inside/outside) Only one voxel is taken into account, but its position is shifted along the normal to the mesh (either inside the mesh or outside), for each mesh vertex (Do not use <Step>).
Line to point Information is taken along the normal line, both inside and outside, with a sampling (depth and step) specified by appropriate parameters.
Inside line to point The value corresponds to <the_choosen_submethod> value for the interpolation for a inside line localized at <Depth> and for a sampling <Step>
Outside line to point The value corresponds to <the_choosen_submethod> value for the interpolation for a ouside line localized at <Depth> and for a sampling <Step>
Sphere to point A sampling into a sphere (depth and step parameters apply) is used to get locations in the 3D volume
- Submethods: This only applies to interpolation methods that are not single-voxel (such as point to point methods)
Table 7.3. Interpolation sub-methods
Section Description Max The maximum value of all voxels of the volume at the sampled locations is mapped on the mesh
Min The minimun value of all voxels of the volume at the sampled locations is mapped on the mesh
Mean Standard mean (sum of values divided by the number of locations)
Corrected mean Only non-nul values are taken into account in the mean computation: this is more suitable for thresholded activation maps for instance to avoid blurring the mapped values.
Enhanced mean In the enhanced mean variant, a weighting of the final value is applied depending on the proportion of null values in the set of mixed values.
- Parameters: definition of localization of another point to do an interpolation
NOTE: be aware that all this is only a visualization toy and is not very robust: no real interpolation of the volume values is performed to get a continuous intersection along the mesh: especially the methods taking points along normals can produce inaccurate results on high curvature regions (produce discontinuities, map the same voxel value on several vertices etc). The sphere mode is more robust but involves an averaging (blurring) effet, and can take values outside the brain or grey matter ...