3.1 Method
This research proposes a pancreas modeling and
classification method based on statistical shape model
(SSM) and fractal tensor with the theoretical basis of
computational anatomy, fractal theory and tensor
theory. The proposed method mainly consists of three
parts: (1) Pancreas modeling based on the level set
method guided by the SSM; (2) Fractal apparent
modeling of pancreas based on the generalized Ndimensional principal component analysis (GND-PCA)
and the fractal theory; (3) Construction of the classifier
for the pancreas diseases diagnose.
There are four parts in the modeling of pancreas: (1)
Segmentation pancreas accurately from many groups
of abdomen medical images based on human-computer
interaction method under the direction of a doctor, and
construct training samples; (2) Normalized the training
67