Probabilistic Classification of Medical Images Using a Hybrid Neural Network

Abstract

We present a new hybrid neural network to perform the probabilistic classification of medical images. The new classifier produces probabilistic classification with Bayesian confidence measure. Based on the SOM map trained with a large training data set, the proposed hybrid neural network performs the probabilistic classification using PNN algorithm. This hybrid neural network overcomes the shortcomings of the parametric methods, the non-parametric methods, and the SOM method in medical image classification. The proposed classifier has been used to segment the CT sloth data resulting in better numerical results and segmented images.