Esmeralda Ruiz Pujadas1, Martin Buechert1,
Michael Weiner2, Stathis Hadjidemetriou1
1Department of Radiology,
Medical Physics, University Medical Center Freiburg, Freiburg, Germany; 2Department
of Radiology, VA Medical Center, Center for Imaging of Neurodegenerative
Diseases, San Francisco, United States
Image segmentation plays an important role in many
medical imaging applications to evaluate possible diseases in patients. But
mostly medical images contain noise and low contrast and a lot of methods are
being proposed to solve specific problems. Then, our study is based on the
application of normalized cuts, a
general segmentation algorithm, for MRI images. This method is robust to
noise and initialization and it has also been used for medical segmentation
giving promising results. We describe the method and combine it with the
nystrm approximation to reduce the computational cost. Some results are
shown.