Bildanalyse von M-FISH

Multiplex-FISH is a combinatorial staining technique that allows the simultaneous detection and discrimination of all human chromosomes. Using at least five fluorochromes all chromosomes can be uniquely labeled in a combinatorial way and identified by their specific spectral signature. Within this thesis I developed a novel approach for the automated analysis of M-FISH images, yielding robust classification results and allowing the analysis of M-FISH images of different experiments. The method combines spectral information with spatial information to tesselate the image into regions of similar color. Subsequently a cluster analysis in color space and a final classification step are performed to identify the biological targets. This approach is applicable to images of different M-FISH experiments, allowing the analysis of interchromosomal as well as intrachromosomal abnormalities in the genome. It also allows the 3D analysis of M-FISH labeled chromosomes in interphase nuclei.