Abstract
Segmentation of the images obtained from magnetic resonance imaging (MRI) is an important step in the visualization of soft tissues in the human body. The new emerging field of artificial neural networks (ANNs) promises to provide unique solutions for the pattern classification of medical images. In this preliminary study, we report an application of Hopfield neural network (HNN) for the multispectral unsupervised classification of magnetic resonance (MR) images. We formulate the problem as minimization of an energy function constructed with two terms, the cost-term which is the sum of squares errors, and the second term is a temporary noise added to the cost-term as an excitation to the network to escape from certain local minimums and be close to the global minimum. We present results from subjects with normal and abnormal physiological conditions obtained using HNN with two and three channels data segmentation.