Probabilistic segmentation of volume data for visualization using SOM-PNN classifier

Abstract

We present a new probabilistic classifier, called SOM-PNN classifier, for volume data classification and visualization. The new classifier produces probabilistic classification with Bayesian confidence measure which is highly desirable in volume rendering. Based on the SOM map trained with a large training data set, our SOM-PNN classifier performs the probabilistic classification using the PNN algorithm. This combined use of SOM and PNN overcomes the shortcomings of the parametric methods, the nonparametric methods, and the SOM method. The proposed SOM-PNN classifier has been used to segment the CT sloth data and the 20 human MRI brain volumes resulting in much more informative 3D rendering with more details and less artifacts than other methods. Numerical comparisons demonstrate that the SOM-PNN classifier is a fast, accurate and probabilistic classifier for volume rendering.