Multispectral MR images segmentation using SOM network

Author

Qian, Tianbai ; Li, Minglu

Author_Institution

Dept. of Comput. Sci. & Eng., Shanghai Jiaotong Univ., China

fYear

2004

fDate

14-16 Sept. 2004

Firstpage

155

Lastpage

158

Abstract

The precise segmentation of magnetic resonance images (MRI) is an important subject in both medical and computer science communities. With MRI´s property of multi-spectrum, we use the information from its PD-,T1-, and T2-weighted images, mapping them into a multi-dimensional intensity space and getting its vector gradient. Through the improvement of the step function, an unsupervised self-organizing map (SOM) neural network is trained dynamically. To improve the effectiveness of segmentation, we develop a semi-supervised training scheme at the edge of image in multi-dimensional intensity space.

Keywords

biomedical MRI; edge detection; image segmentation; medical image processing; multidimensional signal processing; self-organising feature maps; unsupervised learning; MRI; PD-weighted images; SOM network; SOM neural network; T1-weighted images; T2-weighted images; computer science; image edge; magnetic resonance images; medical science; multidimensional intensity space; multispectral MR image segmentation; semisupervised training; step function; unsupervised self-organizing map; vector gradient; Artificial neural networks; Biomedical engineering; Biomedical imaging; Computer science; Image segmentation; Magnetic resonance; Magnetic resonance imaging; Neural networks; Neurons; Protons;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer and Information Technology, 2004. CIT '04. The Fourth International Conference on

Print_ISBN

0-7695-2216-5

Type

conf

DOI

10.1109/CIT.2004.1357189

Filename

1357189