Attenuation Segmentation

The tomographic data produced by a scan contains vast amounts of information, that benefit from specific tools designed to unpack it for interpretation.

The Attenuation Segmentation toolbox allows the user to inspect and interact with histogram data of the voxel attenuation, to discover trends, mark out regions of interest, and visually segment the tomography based on the constructed model.

The tool is found by picking Toolboxes → Attenuation Segmentation from the menu bar.

Histogram Data

When loading a hole, the Attenuation Segmentation toolbox initially displays a histogram computed for the entire hole over a wide range of attenuation values:

To discern relevant features of the histogram, it is typically necessary to zoom in on a narrower range of attenuation values, as controlled through the μ min and μ max boxes:

The user also controls which depth range the histogram data is to be accumulated over; the current depth range values are shown in the Start and Stop boxes, which are updated based on the currently set Depth control mode:

• Preview (default): The depth range values automatically correspond to the entire visible depth range in the Preview area.
• Segment: The depth range values automatically correspond to the selected segment in the Preview area.
• Manual: The depth range values remain unchanged, unless manually edited by the user.

Three different Histogram update modes are also available:

• Auto (default): Histogram data will be automatically loaded whenever the depth range values are changed.
• Manual: Histogram data will be loaded when the user presses the Load histogram button.
• Calculate: Histogram data will be recomputed directly from the tomographic data when the user presses the Calculate histogram button.

The Manual mode is convenient when the user wants fixate the visualized histogram, while still exploring different depths in the Preview and 3D Tomography areas.

The Calculate mode is provided purely for compatibility with older scan data, which lacks precomputed histogram data to load. This mode will be automatically selected if there is no data to load and cannot be directly selected by the user:

Caution

The following information applies only for the Calculate mode.

The large size of the tomographic data makes the histogram calculation a computationally demanding operation. Depending on the computational setup, it may very well be bottlenecked by CPU power, available memory size and speed, and/or the speed of data access via network or disk reading.

Always start by testing the calculation on a relatively short depth range (up to 0.1 or 0.2 meters), then be ready to patiently wait for the results when extending to longer ones.

Adding Regions

Based on the histogram data, it is likely that the user may want to mark particular regions of interest (ROIs) along the attenuation axis. In the example above, there are three clearly discernable features, each centered roughly at 1.0 Al, 1.25 Al and 1.55 Al. A different example will be explored below, with three features spread over a wider attenuation range.

To define a new region, click the Add button, found to the bottom left, which will bring up a dialog for setting the parameters of the region to be added:

Each region is defined by a rectangular function with the specified height, that stretches from its μ min parameter to its μ max parameter.

Each region also has an associated color, opacity (see Coloring the Tomography below) and label; the label allows the user to annotate any useful information regarding, for instance, why the region was added (e.g. "narrow sharp peak at high attenuation") or what it may represent (e.g. "quartz vein").

When the parameters have been set, the region will be plotted out on top of the histogram:

Caution

Regions are saved as part of the Project; remember to always save your project, to not lose any of the added data.

Adjusting Regions

Once added, regions can be edited either by double-clicking them in the table or by clicking the Edit button when the region is selected.

Any number of regions may be removed by selecting them in the table and then clicking the Remove button.

Tip

If a more graphical approach is preferred over the numerical entering of parameters, it is also possible to adjust the attenutation range of any region by selecting it and then dragging one of the sliders marked as Start or End:

If the Lock checkbox is checked, the two sliders will move in unison, effecively translating the region along the attenuation axis.

By adding and adjusting a number of regions, all features of interest in the histogram can eventually be marked out:

Coloring the Tomography

When one or several regions have been added, they may be used as an alternative approach for coloring the tomography by clicking the Color button, found to the bottom right; this will calculate the opacity-weighted sum of each region's color, as a function of attenuation, to produce color and opacity functions for the tomographic rendering.

The end result is a visual segmentation of the entire volume, based on the underlying segmentation of the attenuation histogram:

Tip

If the Auto checkbox is checked, the Color action will be automatically applied whenever a region is added, changed or removed.

Beware that frequent updates may prove demanding to render and therefore inadvisable on less powerful computers.

While the Attenuation Segmentation toolbox is active, the regular Render Settings will be disabled:

Clicking Inactivate will close the Attenuation Segmentation toolbox and return the Render Settings to its previous state.

The coloring produced with the Attenuation Segmentation toolbox can be saved for later reuse by first clicking the Make preset button, which allows to export the coloring as a regular Detailed Preset.

Tip

By combining the functionality of the Lock and Auto checkboxes, it is possible to practially perform a "sweep" over the attenuation axis for interesting ranges:

By dragging the short-ranged but opaque red function over the attenuation axis, various parts of the tomography will be selectively highlighted (as compared to the "background"), which facilitates the search for e.g. familiar patterns.