The Vessel Tree Segmentation Module

Overview

This module allows you to segment tubular structures (blood vessels, bronchial trees) out of a CT dataset. A segmentation and a 3D-Representation of the structure will be generated. You have the oportunity to correct/extend the performace of the segmentation algorithm manually, and also to remove falsely segmented structures from the generated vessel tree.

VesselTreeSegmentation_overview.png

MITK with the VesselTreeSegmentation Module

This document will tell you how to use this module, but it is assumed that you already know how to navigate through the slices of an image using the multi-widget.

For information about the use of the different parts of this module please refer to Details . If you have problems using this module take a look at the Troubleshooting page.

Details

First of all you have to open the data set on which you want to segment tubular structures and select it (single click) in the Data Storage.

Adaptive Region Growing

AdaptiveRegion_widget.png

Adaptive Region Growing Widget

Use the drop-down list to select the structure you want to segment ( Blood Vessels or Bronchial Tree ). According to the taken choice, you have to specify a seed-point for the segmentation by first pressing the seed point button (the mouse cursor will change its shape to a cross) and then selecting a point in the image, inside of a vessel of selected type. After the seed-point selection press [Run segmentation] to obtain a segmentation preview. Use the slider below the button to adapt the estimated threshold value. Checking the box "3D preview" will allow you to see the volume-rendered segmentation preview, which makes the threshold adapting more comfortable.

Press the "Confirm segmentation" button in order to generate a binary image and the both corresponding surface and vessel tree. You can also re-run the whole segmentation step (e.g. with changed position of the seed-point) every time.

Boost Segmentation

If you want to extend the resulting segmentation, switch to the "Boost Segmentation..." tab.

BoostSegmentation_widget.png

Boost Segmentation Widget

There are two possibilities how to process. Simply pressing the [Boost segmentation] button performs the boosting on all branches of the vessel tree (depending on the image size this can be time expensive). The alternative is to specify the branches which shall get extended. Use the branch selection button (after pressing it, the mouse cursor will change its shape to an arrow) to switch to the selection modus. A double-click on a vessel (in the 3D-view window) selects a vessel. Holding the "Ctrl" key while selecting allows you to select multiple branches at once. When the selection is done, start the boosting by pressing [Boost segmentation]. This step can be repeated an arbitrary number of times.

Crop Vessels

You can also remove falsely segmented structures from the vessel tree, switch for this functionality into the "Crop Vessels..." tab.

CropVessels_widget.png

Crop Vessels Widget

The scissors button switches into the selection modus. You can, similar to "Boost segmentation..." functionality, select the branches by a double-click on the vessel tree. You can select multiple branches by holding the "Ctrl" key while selecting. For removing the selected branches please press the button [Remove branches]. This step can be repeated an arbitrary number of times, as long as a vessel tree is present. If you select the whole tree, it will be completely removed. You can return to the "Adaptive Region Growing..." tab to compute another segmentation (e.g. with new settings).

Manual Segmentation

If the adaptive region-growing segmentation does not segment the whole vessel due to pathologies (deformations, stenosis), use the "Manual Semgmentation..." tab for extending the segmentation on a single vessel. Make sure there is a segmentation computed by the adaptive region-growing algorithm. Otherwise this won't work.

ManualSegmentation_widget.png

Manual Segmentation Widget

There are two obligatory points to be set - the start and the end point. Turn on the crosshair for better navigation through the data set. Use the place point buttons and a click into the image to set the start/end point. For improving this segmentation step it is helpful to set some way points between the start and end point. Again use the place points button to switch the interaction mode and set some way-points into the image by "Shift"+click. When all points are set, by pressing [Generate preview] a segmentation preview is computed. After confirming this preview by pressing [Confirm Segmentation] the results will be joined to the segmentation created in the first processing step.

Troubleshooting

I can not select any branches.
Branch selection works considerably better when zoomed out. Try zooming out so that you can see the whole vesseltree and selecting from that point of view.

Every time I try to work on this big data set I get an error message.
Image processing is a highly memory intensive task. If you are using a 32-Bit system it is possible that your computer can not handle that much data. Try the same process on a smaller picture to see if this might be the cause of the problem.

I started the whole tree boosting and the program just stops.
Boosting on a whole complex tree can take a lot of time on some systems. It is recommended to boost only the branches that interest you.

When I select the 3D preview option the program slows down to a standstill.
This option uses your graphic card to speed up the volume rendering. If you have an old or low end graphic card the speed gain might be very small. If this is the case you should consider to forgo this option as it is not necessary for the segmentation but just a visual aid.

Generated on Sun Nov 28 03:25:46 2010 for Mitk 3M3 by  doxygen 1.6.3