An artificial neural network hierarchy for the analysis of cell data

This paper presents an investigation of the use of artificial neural networks in a hierarchical arrangement for the classification of cell image data obtained from smears. The aim is to distinguish between various types of cells and possibly noncellular material based on one or more distinct feature sets obtained from the image data. The extremely divergent characteristics of the cell data makes this a real world classification problem with no easy solution. The paper focuses on the use of backpropagation and learning vector quantization as the artificial neural network classification algorithms. A methodology for the design of the classification hierarchy is presented and the results of experiments involving cell data from smears are analyzed