Segmentation and Linear Measurement for Body Composition Analysis using Slice-O-Matic and Horos

Ribbon muscle degeneration on a sarcopenia has been shown to be a prognostic factor for cancer survival and treatment success. Composition analysis of the third lumbar region is of interest as it provides quantitative characterization of skeletal muscle and within CT scans provides qualitative information of adipose tissue. The following procedures provide protocol and segmented vertical in the sliceOmatic software as well as your measurement protocols for composition analysis in CT and MRI images.

The selection of the L3 vertebrae protocol is completed within GE centricity. Be aware that if using a different software, the tools may vary. Open the patient's CT or MRI images done within the timeframe of interest.

Select monitor regions, 2V window view. Select the coronal or sagittal view on the left screen by navigating to series then selecting the coronal or sagittal images. Select the axial view on the right screen by navigating to series, then selecting the axial images.

Click on conn, cross-reference to link the two images, then scroll through the images on the coronal or sagittal view to ensure the spine vertebrae are easily visible. Next, scroll through the axial view images to find the L1 vertebrae. The L1 is the first vertebrae to not have a rib attachment.

Count from the L1 down to the L3 vertebrae. Be sure to use the sagittal or coronal view for reference to locate the center of the L3 vertebrae. Export the L3 image by navigating to the exam menu, select export images, selected images.

Ensure the format is DICOM. Then select the location the image should be saved to. Double-click on the DICOM file to open up sliceOmatic.

Once the program has loaded, drag the DICOM image anywhere into the sliceOmatic screen. Go to modes, region growing to begin segmentation. Select tools, tag lock.

This feature will be used in subsequent steps. Click on the red one under region growing to set the H.U.limits for muscle. Then click on the off button next to the lower limit.

This will set it to on. Adjust the limit with the slider to be as close to negative 29 as possible. Then use the mouse wheel to set it exactly to negative 29.

Next, click on the off button next to the upper limit to set it to on. Use the slider and mouse wheel to set the H.U.limit to positive 150 Next, click on the green two to set the H.U.limits for I.M.A.T.and set the lower and upper limits the same way but setting the lower limit to negative 190 and the upper limit to negative 30. Click on the yellow five to set the limits for V.A.T.this time set in the lower limit to negative 150 and the upper limit to negative 50.

Finally, click on the cyan seven to set the H.U.limits for S.A.T.setting the lower limit to negative 190 and the upper limit to negative 30. Press the plus and minus keys on the keyboard to zoom the image as necessary. Adjust the zoom as needed throughout the segmentation process.

Select the red one and be sure the brush option is set to paint. Use the brush size that is most efficient for the image and begin painting by clicking on muscle tissue. Paint over the psoas, paraspinal muscle groups, and oblique muscle groups.

Take care on the edges of muscle that are close to spine, internal organs, or fluids, to be sure these are not tagged as muscle. If anything is falsely tagged, select none, and erase the mistake. Once all muscle has been tagged, select one in the tag lock menu.

This will ensure no muscle is accidentally erased or re-tagged as segmentation proceeds. Select the green two and paint over the I.M.A.T.tissue within the muscle fascia. Note, most of the gaps within the muscle itself will be I.M.A.T.Note that the edges of muscle fascia are often lighter in appearance than surrounding tissue.

This can be a helpful guide, particularly in the regions where the oblique muscle groups meet the paraspinal muscle groups and in the region below the erector spinae but above the fascial line. Once all I.M.A.T.has been tagged, select the two under the tag lock menu. Next select the yellow five from the region growing menu to begin segmenting V.A.T.One can still use the brush option, but for many images, it can be easier and faster to use the grow to D option instead.

If using grow to D, be sure to select the smallest brush size, then click on the V.A.T.tissue within the image to tag it. Make sure the tissues or fats inside the limit of intestines or other organs are not tagged as V.A.T.Once all the V.A.T.is tagged, select five from the tag lock menu. Finally, select the cyan seven from the region growing menu to tag S.A.T.tissue.

Again, depending on the image, S.A.T.tissue is often easier to tag with the grow to D option. Once S.A.T.tissue is tagged, check over the edges, particularly around the skin, to be sure only subcutaneous fat is marked as S.A.T.Once all tissues have been tagged, look over the image to be sure there are no errors or mis-marked tissue. When complete, go-to tools, tag surface volume.

Click on the window values button and record each tissue's surface area and average H.U.values. Finally, go to file, save tag files. This will save a tag file where the DICOM image is located.

To begin MRI segmentation, go to modes, region growing. Looking in the preview window, we can see a histogram of H.U.values, where the first peak is air, the next is skeletal muscle, and if the following peaks are discernible, they will be bone followed by adipose tissue. Note, when using MRI images, usually only skeletal muscle can be confidently segmented due to poor adipose tissue discrimination.

Select the red one and turn the lower limit on leaving it set to zero. Then turn the upper limit on and change the mouse wheel option to upper limit. Next, select paint and a large brush option, then move the cursor over the psoas muscle so that the cursor region includes both muscle and bone.

To be confident bone is not tagged as muscle, set the upper limit conservatively by moving the mouse wheel until the upper limit does not include any of the outer bone tissues. Then segment the muscle tissue immediately adjacent to the spine. Next, move the cursor to the linea alba or anywhere else on the scan where adipose tissue and muscle are easy to differentiate.

Increase the upper limit as high as possible without including any adipose tissue. From here, one can proceed segmenting all remaining skeletal muscle. Note, for some images, the upper limit may need to be continually adjusted.

Note, while setting the upper limit with this two-step procedure is conservative, it will optimize the amount of skeletal muscle to be tagged without having to worry about significant erroneous tagging of bone or adipose tissue. When complete, go-to tools, tag surface volume, and select window values to record the surface area and average H.U.values of the skeletal muscle. Linear measurement of the L3 vertebrae region has been shown to closely correlate with total muscle mass.

The advantage to linear measurement is its ease and speed as well as the fact it can be done on any software with a ruler tool or box tool. This protocol is demonstrated using the Horace Medical Imaging app. If using a different software, be aware that the specific tools may very.

Import the image into the viewer by clicking on import then navigating to the image location and selecting open. Once imported, click on the image or patient name to display the image at the bottom. Double-click the image to maximize it, which will also open the tool options.

Identify the psoas and paraspinal muscle groups. Note, the quadratus lumborum should not be considered when determining the edges of the paraspinal muscle groups. Select the ruler tool.

Draw a vertical line from the highest points of the muscle group to the very bottom of the muscle group. Then draw a horizontal line from the widest edge of the muscle group to the other widest edge of the muscle group. Repeat this process for both the right and left psoas and the right and left paraspinal groups.

Note, the lines drawn must intersect at 90 degrees. Note, if the edges of the lines drawn were to be connected, that connection should create a box encompassing the entirety of the muscle group. In the Horace Medical Viewer, it is difficult to use the ruler tool to ensure the lines intersect at 90 degrees.

For this reason, the box drawing tool should be used instead. Select the rectangle tool. Draw a box around each muscle group, now taking note that the edges of the box touch the highest and lowest parts of the muscle as well as the widest parts of the muscle.

As long as the box tool displays its vertical and horizontal dimensions, it may be preferable over the ruler tool, since it allows for easy visualization of the muscle boundaries. A properly segmented CT or MRI image should have no muscle tagging beyond the fascial lines, and for CT images, there should be no adipose tagging within the muscle fascia or on the skin. Record the muscle surface area alongside the patient's height to calculate the skeletal muscle index.

For CT scans, the adipose tissue surface areas may be recorded based on the specific research or clinical questions of interest. Additionally, average H.U.values may be recorded for quality control L3 linear measurement should have the ruler or box measurements encompassing the whole of each muscle. Record the horizontal widths and the vertical heights of each muscle as well as the patient height.

Multiply the widths and heights for each muscle, than sum those areas to obtain the total muscle area. Calculate the linear measure index by dividing the total area by the patient height squared. The demonstrated protocol allows clinicians and researchers to use L3 CT or MRI scans to obtain rapid and reliable quantitative information for the assessment of sarcopenia.