Digital processing of cytogenetic images for their classification

The development of a bioinformatics tool to detect DNA nucleoids (with heads and tails) in fluorescence through image recognition, for their subsequent classification according to the comet assay technique, is presented. The aim is to automate and manage their storage, and to optimize the processes in the cytogenetics laboratory. For this purpose, possible methods to be applied are evaluated and those that combine mathematical and computational algorithms, neural networks and neuro-fuzzy system are selected. The developed method performs the detection and segmentation in the following steps:Pre-processing and initial segmentation of the raw image is carried out. The obtained fragments are classified by neural networks into three groups: head, tail and background. Finally the heads or kernels and tails are measured and re-analyzed, classifying them according to their ratio relationship. This process of detection, segmentation and classification was tested using the case study with cytogenetic images in comet assay of the Laboratory of General Cytogenetics and Environmental Monitoring of the Institute of Subtropical Biology UNaM-IBS-CONICET.