A novel technique for multilevel optimal magnetic resonance brain image thresholding using bacterial foraging

The present paper proposes a novel optimal multilevel thresholding algorithm for brain magnetic resonance image segmentation. This optimization algorithm, employed for image histogram-based thresholding, is based on a relatively recently proposed evolutionary approach, namely, bacterial foraging. Originally proposed towards the fag end of last millennium, bacterial foraging is emerging as a strong contender for distributed control and optimization. The utility of the proposed method is demonstrated by considering several benchmark brain MRI images. The performance of the proposed algorithm, henceforth called BACTFOR, has been compared with another contemporary, popular artificial life based approach introduced for solving complex stochastic optimization problems, namely particle swarm optimization with linearly varying inertia weight (henceforth called PSO_IW). The results obtained for the benchmark images were quite encouraging as BACTFOR could comprehensively outperform PSO_IW.