A New Automated Method for Microscopy Image Analysis: Curvelet Paradigm as a Framework for Sperm Detection

Abstract

Analyzing sperm behavior in semen microscopy images is a modern approach for infertility treatment. distinguishing low contrast sperms from other parts of semen specimen is the major bottleneck of this technique. Machine vision approaches are fitting solutions for detection of sperms but they are challenging. In this article a new method is introduced which utilizes nonlinear mapping in curvelet framework to detect sperms in microscopy images. The proposed method may detect sperms despite of their poor contrasts and vague distribution, thanks to its better sparse representation and more directionality feature than existing approaches. Furthermore, adapting the parameters of the nonlinear mapping due to curvelet components is effective for reinforcement weak ridges as well as better compatibility with different microscopy images. The obtained results show the proposed method achieves the detection rate minimally 4 and maximally 17 percent better than its alternatives, in presence of zero false detection. Furthermore, it is shown that better detection of sperms by proposed method not only does not lead to extract more false objects but also may improve false positive rate by extent of [3-33] percent compared to other examined algorithms.