Module: Multi-Thresholding ()

Description:

The Multi-Thresholding module provides a simple threshold segmentation algorithm applicable to CT or MR image data. The method is also suitable for binary segmentation of other gray level images. Up to five different regions separated by four different thresholds can be extracted. Two regions Exterior and Inside are predefined. However, the name of these regions as well as the corresponding thresholds may be reset by the user.

In order to find suitable thresholds an image histogram showing the (absolute) number of occurrences of voxel values and the current partitioning of the gray value range into the segments is provided. In this histogram several peaks can be identified more or less clearly, The corresponding threshold or segment boundary should be set at the minima between successive peaks. The histogram window pops up by clicking on the Histo action button, adjustments made by moving the value sliders are shown (almost) immediately. You may change the histogram layout (scales, plots of lines and curves, colors) using the Object Editor which pops up when you select Edit Objects in the Edit menu of the histogram window. For more details please refer to the Plot Tool description.

Segmentation starts when you click on the Apply button. Before doing so several options may be set. These options are described in detail below. The segmentation results are stored as a Label Field data object. You may use the Segmentation Editor to further process this object.

The Multi-Thresholding can work with Uniform Scalar Field, however Multi-Thresholding doesn't support Uniform Scalar Field with float values.

Connections:

Data [required]
Image data to be segmented.

Ports:

Regions

This port lets you specify the names of the different regions to be segmented. Between two and five regions may be specified. The individual region names must be separated by blanks.

Exterior-Material1

Lets you set the threshold separating the first and second region.

Material1-Material2

Lets you set the threshold separating the second and third region.

Material2-Inside

Lets you set the threshold separating the third and fourth region.

Options

Toggle subvoxel accuracy causes certain weights to be computed, indicating the degree of confidence of the assignment of a voxel to a particular region. This information is used by the surface reconstruction algorithm to create smooth boundary surfaces, see the Generate Surface module. If no weights are present quite blocky surfaces occur. Note that weights can also be defined using the smoothing filter of the Segmentation Editor.

The second option remove couch will detect the biggest connected component of voxels not assigned to the first region, i.e., Exterior and assign them to the Exterior.

Finally, if option remove bubbles is set an algorithm similar to remove couch is applied in order to detect bubbles inside the Exterior. Because of their low intensity values otherwise these regions would be assigned to Exterior.

Action

The Histo button makes a window pop up showing a histogram for the input image data.