Skip to main content
Fig. 3 | BioMedical Engineering OnLine

Fig. 3

From: Automatic liver segmentation based on appearance and context information

Fig. 3

Illustration of the training procedure of prior liver model. Given training data and the corresponding segmented liver (a), the first classifier is learned based on appearance features and the initial probability distribution map (b) is obtained. Then the appearance and context features are combined to learn a subsequent classifiers (c, d). Here, the context features are extracted from the probability distribution map produced by the previous classifier. After U + 1 iterations, the final probability distribution map \({\mathbf{P}}_{U} \left( {x_{t} } \right)\) is realized (e)

Back to article page