Analysis of multi-fluorescence signals using a modified Self-Organizing Feature Map

Abstract

This paper introduces an algorithm for the multi-sensory integration of signals from the fluorescence microscopy. For the cluster analysis a Self-Organizing Feature Map (SOFM) is used. One basic property of these artificial neural nets is the smoothing of the input vectors and thus a certain insensitivity to clusters of low feature density. While classifying clusters of highly different feature density this property is undesirable. A modification of the learning algorithm of the SOFM, which makes a reproduction of low feature density clusters on a SOFM possible, is described.