ICSI/IVF treatments allocation using CASAs compared to manual semen analyses

The manual method of semen analysis is vital in andrology laboratories. To solve the labor-intensive, time-consuming, and subjective problem, computer-assisted sperm analysis (CASA) systems have been developed. However, it is unclear whether the consistency of semen analysis results especially in male infertility diagnoses can be achieved. A total of 326 individuals were recruited from January 14, 2020, to October 22, 2020. The manual method was used as the gold standard. Pairwise comparisons were conducted between CASAs (Hamilton-Thorne CEROS II Clinical, LensHooke X1 Pro, and SQA-V Gold Sperm Quality Analyzer) and manual method. LensHooke had the best performance in concentration, while the others showed moderate performances (intraclass correlation coefficient [ICC]: CEROS-0.723, LensHooke-0.842, SQA-V-0.631). CEROS II had moderate performances in motility, and the others only had poor agreements (ICC: CEROS-0.634, LensHooke-0.417, SQA-V-0.451). Morphology analyses were not consistent with manual results (ICC: LensHooke-0.160, SQA-V-0.261). In Bland-Altman plots, CEROS II (P = .379) and SQA-V Gold (P = .218) showed consistent measurements in concentrations and total sperm count with the manual method, while the others were inconsistent. LensHooke X1 Pro (κ=0.701) and CEROS II (κ = 0.664) showed substantial performances in oligozoospermia, and SQA-V Gold had moderate agreements (κ = 0.588). In asthenozoospermia, LensHooke X1 Pro had moderate performances (κ = 0.405) and CEROS II had fair agreement (κ = 0.249), while low agreement using SQA-V Gold (κ = 0.157). In teratozoospermia, LensHooke X1 Pro (κ = 0.177) and SQA-V Gold (κ = 0.008) could not have consistent results either. When choosing treatment based on morphology, the ratio of intracytoplasmic sperm injection (ICSI) approximates 0.5 in our unit. However, the ratios were around 0.31 and 0.15 using LensHooke X1 Pro and SQA-V Gold, indicating the reduction of ICSI work in routine treatment. CASA results were not consistent with manual results, and the deviations might result in skewed in vitro fertilization/ICSI allocation in subsequent treatment. Interestingly, tested CASA systems tend to skew to conventional in vitro fertilization instead of ICSI. Although CASA technologies have been improving recently, the manual method cannot be replaced by the tested CASA systems at present and the results should be treated with caution. CASA algorithms should be improved, especially in morphology. Future strict studies should be designed to evaluate the CASA systems with both internal and external validations.