Abstract

Background: A complex network has been studied and applied in various disciplines. As network analysis and image processing are based on matrices, this research analysed the changes in the chromatin image of lymphocyte nuclei in peripheral blood of humans using a network motif and static features (static parameters), so as to complete image classification with network method.\n\nMethods: Image processing technology was used to establish a chromatin image network of a cell nucleus; Network analysis tool Pajek was used to display the special motif of an isolated structural hole with different symmetric line values; afterwards, the frequency of occurrence of this structural hole in patients with nasopharyngeal carcinoma and AIDS, and healthy people was computed. Then by applying the network static features as variables, the chromatin images of stained lymphocytes from the three groups of people were classified and recognised by using an extreme learning machine (ELM).\n\nResults: The frequency of occurrence of the isolated structural hole with different symmetric line values was adopted to distinguish the structures of the chromatins of peripheral blood lymphocytes in patients with nasopharyngeal carcinoma and AIDS, and healthy people. Similarly, The static features of the chromatin image network of a cell nucleus were applied to classify and recognise the morphological and structural changes in chromatins for peripheral blood lymphocytes in the three groups of people.\n\nConclusion: The surface chemical and physical characteristics, as well as the polymerisation link status of biomacromolecules such as DNA, RNA, and protein in the lymphocyte nucleus change under certain pathological conditions. The change influences the combination of small molecular staining materials and any associated biomacromolecules. Therefore, various macroscopic and microscopic changes were found in the chromatin images of the cell nucleus. The microscopic changes include the variations of the extent of staining of chromatin in the nuclei, coarseness and direction of the texture therein, the size of stained conglomerations, etc. These changes contribute to the differences in chromatin image networks among the same type of cells across the three groups. Based on this, the model can be used to classify and reorganise certain diseases. The results prove that using complex network to analyse the chromatin structure of a cell nucleus is of significance.