THE ARTIFICIAL NEURAL NETWORKS IMPLEMENTATION IN GLIOBLASTOMA DIAGNOSTICS

摘要

The results of classification of gene expression profiles in human brain tumor — glioblastoma — and healthy tissue are presented. It is shown, that difference in gene expression profiles is significant enough to make it possible diagnostics implementing artificial intelligence methods. Self-organizing maps and perceptron were used in the study. In the first case we have got a good discrimination of gene expression profiles on the trained map. In the second case trained artificial neural network did perform a successful classification on the test data set — 97.7% were classified correctly.