Automatic lung CT clustering using MKM Algorithm Optimized by PSO

Automatic detection systems in medical sciences improve the accuracy of the diagnosis and reduce the time of analysis. This field in lung disease referred to computer aided diagnosis (CAD) system. Lung nodule detection is the most important task of these systems. CAD systems used the combination of image processing techniques in order to detect the lung nodules. Lung CT image clustering is one of the most important steps in CAT systems. The accurate clustering method reduces the complex following steps and reduces the rate of fault. In this paper we propose an improved MKM clustering method which resolves the weakness of the previous methods and provides an accurate segmentation. In conventional concept of moving, the members in cluster with highest fitness are forced to move to the cluster with the lowest fitness. In the enhanced version of the Moving K-Means algorithm the members move in a defined range. We use particle swarm optimization (PSO) algorithm to optimize the range of the moving radius. Our proposed optimization algorithm with PSO clusters the CT images into the homogenous segment and resolves the problem of the dead center and the trapped center. Simulation results show the effectiveness of the proposed algorithm. The resultant of moving radius range shows that the proposed algorithm fulfilled the relationship condition with optimum value of R=1.8998 e−4and (R) ́=5.6045 e−4.