New Method For Classification Of Spermatozoa Morphology Abnormalities Based On Macroscopic Video Of Human Semen

Two factors that cause male infertility include spermatozoa and nonspermatozoa factors. To discover infertility caused by spermatozoa factor, sperm analysis can be conducted. One of the analysis processes is analyzing the morphology (that is related to the shape of head, neck, and tail) of spermatozoa. The analysis of sperm abnormality is based on the current morphology by observing and calculating the normal shape of sperms manually. Therefore, a system to analyze spermatozoa automatically is required so as to provide essential information regarding sperm condition accurately and quickly. The contribution of this research is to classify spermatozoa automatically and distinguish objects that are not spermatozoa. The novel element of this research is the analysis of sperm morphology based on microscopic videos of spermatozoa by comparing it with the previous researches with images as input data. Hence, what was conducted in this research is automation in analyzing spermatozoa with microscopic video or moving images of spermatozoa as the input data. The research method was initiated with a preprocessing procedure, which included extracting video into images, erosion and dilation operations. It was followed by the process of object segmentation and detection based on edge line, color, and color intensity using contour detection, feature extraction based on eccentricity and Equivalent Circular Diameter (ECD), and finally, classification using decision tree with accuracy result of 83.81%.