Ahmed AFIFI, Toshiya NAKAGUCHI, Norimichi TSUMURA, Yoichi MIYAKE
Graduate School of Sience and Technology,
Chiba University
1-33,Yayoi-cho, Inage, Chiba, 263-8522,
MEDICAL IMAGING TECHNOLOGY Vol.28 No.1 January 2010.
Abstract
Accurate medical diagnosis requires the segmentation of a large number
of medical images. Although manual segmentation provides good results,
it is a costly process in terms of both money and time. Automatic segmentation,
on the other hand, remains a challenge due to low image contrast and ill-defined
boundaries. In this report, we propose a fully automated medical image
segmentation framework in which the segmentation process is constrained
by two prior models: a shape prior model and a texture prior model. The
shape prior model is constructed from a set of manually segmented images
using principal component analysis (PCA), while wavelet packet decomposition
is used to extract the texture features. The Fisher linear discriminant
algorithm is employed to build the texture prior model from the set of
texture features and to perform preliminary segmentation. Then, the particle
swarm optimization (PSO) algorithm is used to refine the preliminary segmentation
according to the shape prior model. In this work, we tested the efficacy
of the proposed technique for segmentation of the liver in abdominal CT
scans. The obtained results demonstrated the efficiency of the proposed
technique in accurately delineating the target objects.
Key words: Medical image segmentation, Shape priors, Particle swarm optimization,
Liver segmentation
Full paper(to appeare)
[PDF]