Segmentation and detection of human spermatozoa using modified Pulse Coupled Neural Network optimized by Particle Swarm Optimization with Mutual Information

In medical imaging field, detection of sperm in sperm images are important in classifying male infertility cases. However, in some cases, analysis of sperm images shows much wrong detection due to poor image quality and multiple target objects. Thus, this study presents a method of image segmentation and detection technique in human spermatozoa image using a modified Pulse Coupled Neural Network (PCNN). As comparison to conventional PCNN, the modified PCNN is proposed with less number of parameters. Although number of parameters is reduced, the proposed method still has difficulty on choosing parameters value. So, the network is optimized with Particle Swarm Optimization (PSO) where a new fitness function was introduced as Mutual Information. Utilizing modified PCNN in such an application is not reported in any literature before. Besides that, this paper also applies Laplacian of Gaussian (LoG) filter on sperm images to detect the centroid of human sperm heads. Qualitative and quantitative assessments show higher accuracy and precision in detecting sperm feature than current existing sperm segmentation method namely Abbiramy's method.