Lesion Segmentation

If you need to go beyond maximum SUV measurements (which UniSyn MI does, of course) and measure mean lesion SUV, metabolic volumes or lesion glycolysis then you'll need to be able to segment FDG avid lesions. UniSyn MI supports two approaches for segmenting molecular imaging data: manual and threshold-based segmentation.

The manual approach is achieved by delineating regions-0f-interest using 2D ROI tools on adjacent slices of the image volume. Once there is at least two 2D contours, UniSyn MI will calculate the volumetric measurements for the 3D region-of-interest. The 3D statistics will get update for every additional 2D contour belonging to the same region. When dealing with larger structures, you may be able to skip contouring on several slices and fill in the blanks using interpolation.

For well defined foci, UniSyn MI's semi-automated segmentation tool will make the task much faster and reproducible. This tool relies on user-defined intensity thresholds to perform a 3D segmentation. Threshold can be specified as a percentage of the maximum value if a 3D volume-of-interest (VOI) or as an absolute SUV value (for PET and quantitative SPECT series).  The segmentation routine will also generate a long-axis (LA) measurement of the segmented region on the slice where the LA is largest. The resulting set of contours can be manually edited, if required.

UniSyn MI only supports threshold segmentation of on the original transverse slice planes, but the segmented contours can easily be rendered on the other MPR views for display purposes.