Detection of compact low-chromatin areas in cell nuclei images

When analysing some DNA stained human cell nuclei using a light microscope or an quantitative image cytometer, compact low-chromatin areas (CLCA) can be observed. We are still not certain about the meaning and source of this phenomena. To enable the detection of CLCA by an image cytometer, a special image processing algorithm has to be developed and new nuclear cell features have to be designed. The presented image processing algorithm automatically detects CLCA in nuclei cell images taken from different tissues. The algorithm is composed of many basic image processing operations. The sequence of operations is determined by a priory knowledge about the properties of CLCA as a set of heuristic roles. The calculated CLCA features are CLCA area, perimeter, average intensity, compactnes and edge sharpness. The matching of the automatic CLCA detection and manual detection (performed by a biologist) was tested using 1400 cell nuclei. The results show a 83.4% match for the nuclei without CLCA and a 93.8% match for the cells with CLCA.