Automated Sperm Assessment Framework and Neural Network Specialized for Sperm Video Recognition

Infertility is a global health problem, and an increasing number of couples are seeking medical assistance to achieve reproduction, at least half of which are caused by men. The success rate of assisted reproductive technologies depends on sperm assessment, in which experts determine whether sperm can be used for reproduction based on morphology and motility of sperm. Previous sperm assessment studies with deep learning have used datasets comprising images that include only sperm heads, which cannot consider motility and other morphologies of sperm. Furthermore, the labels of the dataset are one-hot, which provides insufficient support for experts, because assessment results are inconsistent between experts, and they have no absolute answer. Therefore, we constructed the video dataset for sperm assessment whose videos include sperm head as well as neck and tail, and its labels were annotated with soft-label. Furthermore, we proposed the sperm assessment framework and the neural network, RoSTFine, for sperm video recognition. Experimental results showed that RoS-TFine could improve the sperm assessment performances compared to existing video recognition models and focus strongly on important sperm parts (i.e., head and neck). Our code is publickly available at https://github.com/FTKR12/RoSTFine.

1. Introduction

Infertility is a critical problem around the world. This afflicts one in six couples, at least half of whom are casued by men [19], [16]. Assisted reproductive technologies (ARTs), such as in-vitro-fertilization (IVF) and intracytoplasmic sperm injection (ICSI), are used depending on the cause and severity of infertility. However, ARTs are currently successful in only approximately 33% of cases, and this main reason is suboptimal sperm selection [27]. In the sperm selection process, at least three fertility factors are typically examined; sperm concentration, motility and morphology [20]. In sperm selection, motility and sperm concentration are assessed using computer-aided semen analysis (CASA) systems, which are sensitive to sample preparation and equipment setup [37], [2]. Morphology is assessed manually by experts, which are inconsistent among individuals and clinics owing to subjective criteria, in addition to being time-consuming and labor-intensive [13], [8], [25], [22]. Therefore, an End2End sperm assessment framework, considering all three factors, is in high demand and promising for improving reproductive success.