Mask Generators panel

This panel hosts several ways of automatic mask generation. Specific areas of interest from the generated mask may further be selected for segmentation. See more about masks.

Back to Index --> User Guide --> Panels

Contents

Common fields

There are several common fields that do not depend on type of the selected mask generator.

the whole dataset in the 2D mode, i.e. slice by slice; 3D use the 3D mode for the mask generation

3. Do it button:

Frangi Filter

Hessian based Frangi Vesselness filter, written by Marc Schrijver and Dirk-Jan Kroon. This function uses the eigenvectors of the Hessian to compute the likeliness of an image region to contain vessels or other image ridges, according to the method described by Frangi 1998, 2001.

Note, to work properly this function should be compiled. See details in System Requirements

Parameters:

Morphological filters

Set of Matlab based morphological filters.

Strel Filter

Generate mask based on morphological image opening and black-and-white thresholding. The function first performs morphological bottom-hat (Black on white is checked, 5.) or top-hat (Black on white is unchecked, 5.) filtering of the image. The top-hat filtering computes the morphological opening of the image (using imopen) and then subtracts the result from the original image. The result is then black and white thresholded with parameter in the B/W threshold edit box (3.).

  1. Strel size, defines size of the structural element (disk type) for imtophat and imbothat filtering.
  2. Fill checkbox, check it to fill holes in the resulted Mask image.
  3. B/W threshold, specifies parameter for the black and white thresholding.
  4. Size limit, limits the size of generated 2D objects so that objects smaller than this value are removed from the Mask during the filter run.
  5. Black on white checkbox, when checked, the filter will use morphological bottom-hat filtering (imbothat). When unchecked - morphological top-hat filtering (imtophat).

BW Threshold Black Filter

Different approaches for black-and-white thresholding. The data should be as bright object on dark background. The thresholding is applied to ROI regions or to the whole image. The resulting mask may be further filtered with tools in the Mask menu.

  1. For bright object extraction, use the 1. editbox to specify a threshold value. The objects that are brighter than the specified value will be used to generate the Mask layer. The auto checkbox 3. should be unchecked!
  2. For dark object extraction, use the 2. editbox to specify a threshold value. The objects that are dimmer than the specified value will be used to generate the Mask layer. The auto checkbox 3. should be unchecked!
  3. Auto checkbox, enables automatic image thresholding using Otsu's method. Note! The auto mode is affected by the values in the 1. and 2. editboxes. If the threshold coefficient is smaller than the parameter in the 1. editbox, the specified 1. parameters will be used instead of automatic coefficient.
  4. Auto grid checkbox, another approach to the automatic thresholding. In this method the image/ROI region is divided into a separate blocks. The program calculates threshold values for each block and use these values to prepare the image for final thresholding. This method works for cases when the image intensity is constant in each block in Z-direction. The grid will be displayed after the run as the Selection layer and can be removed with Ctrl+C / C shortcut.


It is recommended to run grid mode with 0 coefficient in 5.. The image will be somehow thresholded and the used coefficient will be shown in the Matlab command window. If result is not good enough, a new coefficient that is differ from the automatically determined value should be placed into the 5. editbox.

6. Manual grid checkbox, in this mode the coeficients for each box may be manually tuned.


The coefficients may be saved and loaded for future use.

Parallel processing

Code is optimized for Parallel processing. The grid run mode gives x5-6 times better performance with 8 cores. Parallel processing should be enabled by pressing the Turn on parallel processing button in the im_browser toolbar.

Back to Index --> User Guide --> Panels