The Application of Cellular Automaton in Medical Semiautomatic Segmentation - Cellular Automata in Image Processing and Geometry - page 157

Chapter 9
The Application of Cellular Automaton in
Medical Semiautomatic Segmentation
Yonghui Gao and Jie Yang
Abstract. Efficient image segmentation is the basic premise of the medical diagno-
sis and analysis. However, this task exits several major challenges due to the special
anatomy and topological changes. Since fully automated techniques can not guaran-
tee the efficiency and precision in general clinical applications, interactive scheme
is often alternatively performed with a special or multiple algorithms. To evaluate
the performance of an interactive segmentation, three indicators should be consid-
ered: interactive efficiency, implementation complexity and accuracy. An efficient
solution should minimize the interaction and the implementation complexity under
the premise of ensuring higher segmentation accuracy.
This chapter describes an interactive segmentation framework based on Cellu-
lar Automaton (CA). User interaction is performed in the selected image planes to
indicate the position and main features of the object and background. The models
are extracted efficiently by a Cellular Automaton evolution on homogeneous cells
- pixels (voxels), regions and blocks with the corresponding attacking and merging
mechanism. CA based strategy does not consider the marker or non-marker state of
cells, which grants it with the flexibility in parallel implementation and the ability
to deal with multi-class problem. It is easy to implement because only some sim-
ple operations are needed. It is accurate because a more general linear or nonlinear
model can be learned.
Yonghui Gao
University of Shanghai for Science and Technology,
School of Medical Instrument and Food Engineering,
Shanghai, China
e-mail: [e-mail]
Jie Yang
Shanghai Jiaotong University,
Institute of Image Processing and Pattern Recognition, China
e-mail: [e-mail]