Performance evaluation of bio-inspired optimization algorithms in resolving chromosomal occlusions

This work evaluates the performance of bio-inspired optimization algorithms in resolving occlusion in chromosomal images. The presence of occlusion hinders accurate identification and classification in automatic karyotyping and a manual intervention is needed to complete the procedure. For this reason, karyotyping is not completely automatic and a novel technique based on bio-inspired optimization algorithms is proposed to identify the individual chromosomes even in the presence of occlusion. The technique employs stochastic search algorithms including the Firefly algorithm (FA), Genetic algorithm (GA) and Particle swarm Optimization (PSO) in resolving occlusion, by starting with a random population of solutions from the image of occluded chromosomes and recursively doing operations borrowed from evolutionary methods and swarm intelligence, on the population. The hidden chromosomes are identified after a certain number of iterations. The technique performs well, even when 80% of the chromosome is occluded by the other. The performance of the stochastic search algorithms in resolving chromosomal occlusions is evaluated and FA gives superior results in identifying the occluded chromosomes.