Connecting

First, set up the Fuzzy Connector tool as described in the introduction. Next, select any of the following options:

  • fuzzy_combine If selected, all seeds pixels in all defined ROIs will be used as a single seed for fuzzy connections. Statistics about the seed pixels in all ROIs will be combined to characterise the intensity statistics of the feature. If not selected, each ROI will be treated individually, with intensity statistics characterised only for the pixels within that ROI, and fuzzy connectedness being found for each ROI separately. Thus, when this option is not selected, multiple fuzzy connectedness algorithms are run sequentially: one for every seed ROI.
  • fuzzy_connect_3d When selected, connections are considered in three dimensions (in-plane and between slices). When not selected, only in-plane connections are considered. Use this option when the feature you want to find extends across multiple slices, but you do not want to define seed ROIs in all slices.
  • fuzzy_fill_holes When selected, "holes" within features will be filled. When not selected, a hole in a feature (a part within the feature that has a low fuzzy connectivity) will be delineated as a region of interest within another region, as illustrated in the figure below:

fuzzy_hole

Filling holes simply removes all resulting ROIs that are contained within another ROI.

Clicking the save_settings_button button will save the setup so that next time the Fuzzy Connector tool is started, it will have the same setting as in this session.

When you are ready to find the image feature, click the connect_button button. A dialog will pop up to show that the Fuzzy Connector is working:

fuzzy_progress

When the Fuzzy Connector is finished, it will create a set of ROIs that surround the fuzzy-connected feature that contains your seed ROIs. These ROIs will either be:

  • Loaded into the displayed image, if you have selected fuzzy_load_rois. If you are not happy with the Fuzzy Connector's result, you can simply select Undo from the ROI Toolkit's Edit menu (or type CONTROL-Z) to undo the addition of the resulting ROIs, change the Fuzzy Connector's settings, and try again. OR
  • Saved to an ROI disk file, if you have selected fuzzy_save_rois. A File Chooser will pop up, so that you can select a file in which to save the resulting ROIs when you press the connect_button button.

Note: there is an more interactive mode of operation, whereby fuzzy connections can be made to a single pixel. To try this, start the Fuzzy Connector tool, setup the tool to your requirements, but do not define any seed ROIs. Instead, point to a seed pixel on the image (with a slice selected) and press the 'F' key on your keyboard. Fuzzy Connections to this single seed pixel will be generated and loaded as a set of ROIs to the displayed image. Illustrated below is an example of the use of the Fuzzy Connector. The image being analysed is the T1-weighted head image from the image samples that can be downloaded from this web site. The aim is to extract the lateral ventricular system. The setup used for the Fuzzy Connector is shown below:

fuzzy_setup_t1

Fuzzy Connector setup to find the ventricular system.

Note:

  1. The intensity hint "Darker" has been provided, since the CSF in the ventricles on T1-weighted images is darker than the parenchyma.
  2. The threshold setting of 0.52 was derived by trial and error.
  3. Connections are made in 3-dimension so that we do not have to define seed ROIs on every slice - the Fuzzy Connector will find the 3-D connected ventricular system.

Next is the placement of the seed ROIs. We have provided two seed ROIs: one for left and one for the right ventricle. We have used rectangular ROIs, but any shapes could have been used as long as they were contained wholly within the ventricles.

fuzzy_seeds_t1

Seed ROIs in the left and right lateral ventricles. Seeds were only set in this slice.

Below is the resulting set of ROIs for the ventricular system.

fuzzy_result_t1

The fuzzy connected lateral ventricles. The Fuzzy Connector has outlined this feature as a set of ROIs.

The resulting ROIs can be edited or deleted as you wish.

Note: if you use the ROIs produced by the Fuzzy Connector to calculate a volume (for example the volume of the ventricular system, as found in this example), then you should ensure that the seed ROIs are not part of the displayed ROIs, since the volume of these would be added to the total volume, giving an erroneous result. If you want to calculate volumes, it is probably safest if you save the seed ROIs to disk, and set the Fuzzy Connector to load the seed ROIs from that disk file by selecting "Get seed from seeds_roi_file".