Basic Segmentation Test

• Read the Doxygen help page
• Load an image
• Make sure you are using the MITK "Display" as your main editor window.
• Drag and drop the image from the Datamanager to the Drag and Drop Display
• Open the MITK Segmentation Editor. Window->Show View->MITK Segmentation
• Select the image in the Datamanager
• Click on "start/re-start segmentation" in the MITK Segmentation
• Set a name and a color for the segmetation image
• Click "OK"
• Click on the "Add" button for drawing a segmetation region. The Add tool allows the user to draw a freehand contour on an image slice by clicking and dragging. Once the mouse button is released, the inside of the contour will be marked as belonging to the segmentation (i.e. it is filled). For testing it, you have to click and dragging the mouse around the desired closed region.
• Click on the "Substract" button for removing a set of voxels from a segmentation region. The Subtract tool has the opposite effect of the Add tool: it allows an area within the drawn contour to be marked as background. For testing it, you have to click and dragging the mouse drawing the desired region to remove.
• Click on the "Paint" button for drawing voxel a voxel. For testing it, you can click voxel by voxel, adding it in each click to the segmetation mask, or dragging the mouse.
• Click on the "Wipe" button for removing voxel a voxel. For testing it, you can click voxel by voxel, removing it in each click to the segmetation mask, or dragging the mouse.
• Click on the "Region Growing" button for automatic drawing of a region segmentation following region growing algorithm. The interactive region growing tool performs 2D region growing on the slice clicked by the user, i.e. it selects a connected region around the mouse cursor based on a common gray value interval. Parameters for region growing are all determined by mouse interaction. The initial interval is determined by the first mouse click within the image—the initial value suggested for interval width is 20% of visible gray values (as determined by the current level window settings). A contour in the image always shows the region that would be selected by using the current interval. The user can change the interval width by moving the mouse up or down with the mouse button continually pressed. The region growing tool has a second function to alleviate the common problem of leakage: gray value-based region growing will often leak into neighboring structures with the same gray values through thin connecting areas. If this happens, the user can click inside the leak region and the tool will try to detect and remove the leakage area. To achieve this, a skeletonization of the segmentation is calculated and the algorithm looks for the thinnest connecting “bridge” between the old and new area. If such a bridge can be determined, the leak region will be removed.
• Click on the "Correction" button for correcting a segmentation region. It is designed for small corrective changes and supports interactions that do not require the user to actively choose the Add or Subtract tool or even draw a closed contour. The tool allows the user to draw a freehand line (not a closed contour) and decide on the appropriate action depending on the relation between the drawn line and the existing segmentation.
• Click on the "Fill" button for filling the holes. With the Fill tool, the user can click within an unsegmented region that is completely surrounded by segmented parts (a hole). Such a region will be detected and labeled as segmentation. For testing it, click inside a hole of a segmetation mask, check that it is filled in the selected slice only. Click over a segmetation region for filling all the holes in one click.
• Click on the "Erase" button for removing whole segmetation regions. The Erase tool detects the connected segmented area around the point clicked by the user and marks it as unsegmented. This can be used to quickly remove small spots that are not part of the foreground object (islands). Click over the region and it will be removed.
• Move to a new slice using the mouse wheel to scroll and follow drawing.
• Click on the main image in the Datamaneger to finish the segmentation (inactive it) or start a new one.
• Click on the segmentation's name in the Datamaneger and change a previous segmentation.
• Go to Window->Preferences->MITK Segmentation for changing main settings.

Segmentations are identified by a name and a color, both of which can be changed by way of a context menu in the Datamanager. One segmentation can be marked as the active one, indicated by the highlighted item in the Datamanager. All selected segmentation will be displayed as an overlay on the reference image.

Segmentations can be saved to disk in several file formats by way of a context menu in the Datamanager.

For extended information about this tool see D. Maleike et al. (2009) - Interactive segmentation framework of the Medical Imaging Interaction Toolkit