Medical image segmentation using a contextual-constraint-based Hopfield neural cube

Neural-network-based image techniques such as the Hopfield neural networks have been proposed as an alternative approach for image segmentation and ha ve demonstrated benefits over traditional algorithms. However, due to its a rchitecture limitation, image segmentation using traditional Hopfield neura l networks results in the same function as thresholding of image histograms . With this technique high-level contextual information cannot be incorpora ted into the segmentation procedure. As a result, although the traditional Hopfield neural network was capable of segmenting noiseless images, it lack s the capability of noise robustness. In this paper, an innovative Hopfield neural network, called contextual constraint-based Hopfield neural cube (C CBHNC) is proposed for image segmentation. The CCBHNC uses a three-dimensio nal architecture with pixel classification implemented on its third dimensi on. With the three-dimensional architecture, the network is capable of taki ng into account each pixel's feature and its surrounding contextual informa tion. Besides the network architecture, the CCBHNC also differs from the or iginal Hopfield neural network in that a competitive winner-take-all mechan ism is imposed in the evolution of the network. The winner-take-all mechani sm adeptly precludes the necessity of determining the values for the weight ing factors for the hard constraints in the energy function in maintaining feasible results. The proposed CCBHNC approach for image segmentation has b een compared with two existing methods. The simulation results indicate tha t CCBHNC can produce more continuous, and smoother images in comparison wit h the other methods. (C) 2001 Elsevier Science B.V. All rights reserved.