In this step you will use the "Histogram Thresholding" mode to differemtiate the fat and muscle tissues.
|
|
|
|
|
Snake Tagged MR Scan |
|
Histogram Tagged MR Scan |
What the script did for you:
The "Mode ALL" is enabled, showing all the slices.
The TAG labels have been modified:
The "TAG 1" is renamed "Muscle".
The "TAG 2" is renamed "FAT".
The "TAG 4" is renamed "Bone".
The "Histogram Thresholding" mode interface is open.
The number of desired peaks is set to "2".
The histogram is set to be compute from TAG 1 (was Leg, is now Muscle)
All Mixer bypass are set to "off", so the displayed image is a mixed of "Fat-Water".
The "Color Scheme" tool is open.
The color scheme is set to "Mix".
The "TAG Lock" tool is open.
The TAG 0 (outside) is locked.
The TAG 4 (Bone) is locked.
What to do:
|
The only thing left to do is:
•Click on the "Recompute Histogram Thresholds" button.
•Click on the "Compute" button. |
|
The Interface:
|
The Histogram Thresholding
Using the Histogram Thresholding mode, You will segment the muscle and fat tissues in side the leg.
The basic principle of this mode is rather straight forward: it compute the histogram of all the pixels selected using a certain criteria. It then filter that histogram until it only contain the desired number of peaks. For each of these peaks we have a range of grey level values. Each of these ranges is then assigned to a TAG value. By default, sequential TAGs starting at "TAG 1" are assigned the lower and upper limits of their matching peaks. This can be modified using the "Off/Manual/Auto" options of each tags.
The script already selected the desired number of peaks at "2" and limit the computation of the histogram to only the pixels under the "TAG 1" (Muscle) the tag that we used to segment the leg with the snakes. The histogram has already been recomputed for these values.
The TAG Lock tool has been set to lock tag "None" and "Bone", so that the result of the segmentation will only be applied to the interior of the leg with the exclusion of the bone.
The program now show you a preview of the result of the segmentation. (You can activate/deactivate the preview). The actual computation is only done when you click the "Compute" button.
If you want, you can click on the "Blow" button to see the histogram in a separate window. Clicking on this button twice will make the window "topmost" and prevent it from disapearing behing sliceOmatic. In the histogram window, you can zoom on a section of the histogram by dragging a box over the section you want to see. The histogram window will show you the histogram data (in yellow), the smoothed curve (in magenta) and the first derivative of the smoothed data (in grey). The peaks are found at the zero crossing points of the first derivative. The TAG transition is done at the saddle point between the peaks.
|
|
Note:
|
The Fat and Muscle are defined as tissues under each peak of the histogram. You may be tempted to change the threshold value manually to "get a better fit", but the important thing here is consitency. You will want to study the evolution of the Fat/Muscle ratio. So we let the program decide where the threshold lies. |