Improved Toboggan Segmentation Algorithm for Magnetic Resonance Images
Li Guo, Jianhua Wu, Pian Zhao-yu, Wang Kun · 2007
The precise segmentation of magnetic resonance images (MRI) is an important subject in both medical and computer science communities. The intrinsic complexity of the images and their relative lack of systematics have brought to develop different approaches to segment the different parts of human head. This paper investigates a novel feature extraction approach to MRI segmentation based on feed-back pulse coupled neural network in conjunction with toboggan theory. Due to the dynamics of the FPCNN, multiple unconnected groups of neurons will often pulse at the same time, calling for further processing to identify distinct regions. We locate the object's label by FPCNN. Finally, toboggan automatically partitions the MRI image. The experimental results show that the proposed algorithm performs well compared to the traditional algorithms.