Authors:
Liang Geng
and
Weibei Dou
Affiliation:
Tsinghua University, China
Keyword(s):
Brain tissue classification, Fuzzy MRF Model, multifractal analysis.
Related
Ontology
Subjects/Areas/Topics:
Applications and Services
;
Biomedical Engineering
;
Biomedical Signal Processing
;
Computer Vision, Visualization and Computer Graphics
;
Medical Image Detection, Acquisition, Analysis and Processing
Abstract:
This paper introduces multifractal analysis to the Fuzzy Markov Random Field (MRF) Model, used for brain tissue classification of Magnetic Resonance Images (MRI). The traditional classifying method using Fuzzy MRF Model is already able to calculate out the memberships of each voxel, to solve the Partial Volume Effect (PVE). But its accuracy is relatively low, for its spatial resolution is not high enough. Therefore the multifractal analysis is brought in to raise the accuracy by providing local information. The improved method is tests on both simulated data and real images, where results on membership average errors and position errors are calculated. These results show that the improved method can provide much higher accuracy.