MRI Object Modeling Workbench - User Manual
About MRI Object Modeling Workbench
The OMW allows to extract objects from large datasets. It has been
created for the purpose of 3d-modeling but other usages are possible,
for example
object tracing in time series. It has been used for the
3d-reconstruction of a rat's jaw from histological sections, the
reconstruction of a palm leave from histological sections and the
3d-modelling of parts of a palm meristem from confocal microscopy.

When a series of images is loaded, the slices are listed in a sidepane.
When an image is selected it is loaded and shown. Two images can be
shown in the same time in the same window. The slices can be aligned
automatically, based on the difference between two images for each
alignment position, or manually. When a selection on a slice is made it
is automatically kept. A special magic wand tool for the selection of
compartments is included in the OMW. Objects can be defined by a number
of selections. Finally a stack for each object can be saved to disk.
This stack can be used as the input data for the 3d modeling.
Features:
- stack registration
- transparent overlay of images
- magic wand for compartment selection
- definition of objects from selections
- transparent display of objects
- subtraction and addition of object areas
- export of objects as image data
- all ImageJ commands can be applied to the images in the worbench display
- can be used as a convenient way to create batch processing jobs on image series
Limitations:
- no 3d modeling and display, but a stack image of the objects can
be created in memory and the ImageJ 3d-plugins/commands can be used on
it
- 1 object can only have 1 selection per slice and composite selections are not allowed
- only two images can be displayed in the same time
- the image window only works on the primary monitor
Starting the OMW
Start the object modeling workbench either from Show>object modeling
workbench or by clicking the OMW button on the MRI-CIA taskbar.


Opening an image series
File>open series... will bring up the list editor. Press add to add
a directory or the selected images. If you add a directory all
tiff-images in all subdirectories will be added. Before closing the
list editor you can remove unwanted images from the list. If
your data is in another format than tif-series, you have to
convert it before loading it into the OMW.

Image display
Select a slice from the list at the left hand. The according image will
be loaded and displayed. You can select more than one slice in the same
time. When more than one slice is selected, the first and the last
selecteded slices are displayed and all commands are send to the last
selected slice. With a right click on a slice list entry you can open
the slice's display options. You can change the opacity of a slice. An
opacity of 100 means fully opaque and an opacy of 0 means fully
transparent, thus invisible.
You can apply MRI-CIA operations and ImageJ commands to an image
in the OMW. However when you de-select the image all changes are lost,
except for:
- the lookup-table
- the opacity
- the selection (roi)
The zoom you choose, by zooming in or out, will be used globally for all images.

Automatic stack registration
The images from histological sections have to be aligned before one can
start to extract objects. This is called stack registration. You can do
it either manually or automatically. To start the
automatic registration use Process>>calculate
translations. You will be asked where to save the result file. The
translations file is a text file where each row contains three entries,
seperated by tabs. The first entry is the x-offset, the second the
y-offset and the third the score. The score indicates the quality of
the alignment. The lower the score the better the alignment. It is the
average difference between the two images in the overlapping region for
a given translation.
The computation is finished when the ImageJ progress bar is not moving anymore.


Once the translations are written you can apply them to the opened
image series by using File>open translations... The offset and the
score will then be displayed in the object modeling workbench and the
images will be displayed in the according positions.

The calculation of the translations can take a long time. You might
eventually want to scale down the images before starting the
translation calculation.
Alternatively you can use Process>>calculate translations
quickly. You will be asked for a step width. The images will be
compared using a raster with cells of size step width. The final
position will be searched around the best fit on the raster. Although
this is much faster, it might, depending on your images, create more
false alignments.
When the translations are computed you should look for translations
with a high score. These are likely to be false and you should correct
them manually.
Scaling images
To scale up or down all images in a series use Process>>scale
series... You'll be asked for the x- and y- scale factor and if
interpolation should be used. The you have to indicate a destination
folder for the scaled series. The translations will be scaled
accordingly and the scaled series is loaded after the processing
finished.
Manual stack registration
Click into the image canvas to activate it. You can now use the arrow
keys to move the second selected images in all directions. The movement
will accelerate as long as you you keep a key pressed. The
current offset and the current score are displayed in the MRI-CIA
status bar. Once you found the right translation you can apply it to
the moved slice only by pressing ENTER and to the moved slice and all
following slices by pressing CTRL+ENTER.