J. Horáček, M. Horák, and J. Pelikán (Czech Republic)
Image Segmentation, Medical Imaging, CT, Femoral Head, Dynamic Programming.
We present a new improvement to the segmentation process
of femoral head from CT dataset after a femoral neck fracture. Many existing algorithms for femoral head segmentation are slice-by-slice based and usually have problems
with incorporating information from subsequent slices and
that may lead to discontinuous result between individual slices. We overcome this weakness by performing a
dynamic programming approach on data transformed to
spherical coordinate space. The overall user interface is
much more user friendly, because now the only segmentation parameter type on the user side is a point helping to
localize the corticallis. This paper also discusses how to
help the algorithm over areas where the automatic segmentation may have problems, especially when the data are of
elderly patients with osteoporosis or damaged cortical parts
of the bone.