Classification of acute leukemia sub-types using artificial neural network

Abstract

This paper presents a study on sub-types classification of acute leukemia using artificial neural networks. Thirteen morphologival features have been extracted from acute leukemia cells and used as the neural network inputs for the classification. Multilayered Perceptron networks were used to perform the classification task. Multiclass and hierarchy binary class were used to perform and compared using 2000 data samples. The classification results indicating that the hierarchy classifier gives the better overall diagnostic performance compare to the multiclass classification. The overall accuracy of hierarchy classifier is 96.65% whereas the multiclass classifier overall accuracy is 88.55% for 20 hidden nodes.