The GUI

the thr_ panel

binary images through thresholding (segmentation)

adapted from
PLoS supp. material by Hermann Cuntz

← the_stk_panel
→ the_skl_panel

With the thr_ panel a set of binary matrices is constructed by thresholding the original image stacks. If the „dyn.“ toggle is pressed down a local (or dynamic) threshold is taken. This is particularly advantageous to capture small branches. The edit value is an offset to the local mean brightness necessary to be part of the binary matrix. Without dyn. thresholding the edit value indicates the actual absolute threshold value. The vis_ panel displays the percentage of voxels which fall in the binary matrix. That should typically not exceed 2-3%. When the third radio button is switched on, the maximum intensity projections of the stk_ panel show the index of thresholded maximum values, which in the XY view for example corresponds to the z values (between 0 and 1), at which the brightness is maximal in the original image stack.

With the first radio button switched on, little blue transparent tiles are drawn on the maximum intensity projections to show which voxels will be kept in the binary matrices.

Cleaning the binary matrix involves linking voxels by neighborhood relationships. When a set of linked voxels is smaller than the cleaning size in the edit field, it is removed. This is an efficient way to remove noisy bits off of the thresholded image (see Matlab function „bwareaopen“).

thr_ edit mode
In the thr_ panel edit mode the binary matrices can be manually edited by setting an individual threshold in a marked area. This method allows to change the threshold value in the edit field and then clean the binary matrix locally. The size of the local areas can be altered with the keys [r], [R], [f], and [F]. To switch to a different image stack it is necessary to switch back to the stk_ panel.

Finally, in order to check the integrity of the binary matrix and its flow between the tiled image stacks, the second radio button can be activated (this can be very slow). This shows a 3D box representation of the binary matrices at a reduced resolution (2x2 voxels in the XY plane).