Feedforward Neural Networks for the Classification of Two-dimensional Polyacrylamide Gel Electrophoresis Images

Abstract

This article describes a method, using neural networks, for classifying two-dimensional polyacrylamide gel electrophoretograms, complex biomedical images that contain proteins separated from a biological sample. The classification aims at grouping images and identifying their most significant features. The gel image processing part is first summarized. The details on how the classification is accomplished using neural networks are then presented. After that, an experiment using real gels of rat cells is carried out, showing the successful implementation and application of this method. Finally, experimental results show that this neural network based method is more than 90% effective.

This article describes a method, using neural networks, for classifying two-dimensional polyacrylamide gel electrophoretograms, complex biomedical images that contain proteins separated from a biological sample. The classification aims at grouping images and identifying their most significant features. The gel image processing part is first summarized. The details on how the classification is accomplished using neural networks are then presented. After that, an experiment using real gels of rat cells is carried out, showing the successful implementation and application of this method. Finally, experimental results show that this neural network based method is more than 90% effective.