The segmentation of CT images to produce a computational model of anatomy is a time-consuming and laborious process. Here we report a time saving semi-automatic approach. The image-processing technique known as "statistical region merging" (SRM) was used to pre-segment the 54 original CT images of the ADELAIDE data set into regions of related pixels. These regions were amalgamated into organs and tissues by a program operated through a graphical user interface. This combination of SRM and GUI was used to build a voxel computational model of anatomy. The "new" version of ADELAIDE was compared to the "old" version by simulating an abdominal CT procedure on both models and comparing the Monte Carlo calculated organ doses. Seventeen of the 21 SRM-GUI segmented tissues received doses that were within 18 % of the doses received by the manually segmented tissues. Hence the SRM-GUI segmentation technique can produce a computational model that is not functionally different from a manually segmented computational model. The SRM-GUI segmentation technique is able to reduce the time taken to construct a voxel tomographic model from CT images.
|Number of pages||11|
|Journal||Australasian Physical and Engineering Sciences in Medicine|
|Publication status||Published - Jun 2014|
- CT dosimetry
- Image segmentation
- Statistical region merging
- Voxel model