A comparison of cluster distance metrics for the segmentation of sputum color image using unsupervised hopfield neural network classifier

Unsupervised Hopfield Neural Network classifier (UHNNC) is an operational process appropriate for the segmentation of different type of medical and natural images. Its efficiency subsidizes not only to its start from a random initialization for the assignment of each pixel to only and only one cluster but also to its convergence to an advanced optimal solution in a pre-specified number of iterations. In this paper, we present a study of the distance type effect on the segmentation result using UHNNC. We have used a database of 1000 sputum color images prepared to be used in a screening process for lung cancer diagnosis. A quantitative comparison between the results obtained using the Euclidian and the Manhattan distance or city block distance showed that the former gives better classification or segmentation to the pixels of the different cells present in the sputum color images.