Segmentation of heart in computed tomography images using swarm and evolutionary active contours

This paper presents a new image segmentation scheme based on active contours guided by the optimization techniques Particle Swarm Optimization (PSO) and Differential Evolution (DE), independently. The scheme uses the optimization methods over a polar coordinate system to perform the segmentation task increasing the energy-minimizing capability regarding the traditional active contour model. This proposed model is applied in the segmentation of the human heart from datasets of sequential Computed Tomography images. In addition, to obtain a quantitative and qualitative evaluation of the segmentation results compared to regions outlined by experts, different similarity metrics have been adopted. The experimental results demonstrate that by using PSO or DE, the proposed scheme outperforms the traditional implementation of active contour model in terms of stability and efficiency, achieving a high accuracy segmentation regarding the ground truth.