Seymour Katherine Rose, Rickard Jessica P, Pool Kelsey R, Pini Taylor, de Graaf Simon P
School of Life and Environmental Sciences, Faculty of Science, The University of Sydney, Room 344, RMC Gunn Building, Sydney, NSW, B19, Australia.
School of Agriculture and Environment, The University of Western Australia, Crawley, WA, Australia.
Sci Rep. 2025 Jul 1;15(1):21963. doi: 10.1038/s41598-025-07515-3.
Sperm morphology assessment is recognised as a critical, yet variable, test of male fertility. This variability is due in part to the lack of standardised training for morphologists. This study utilised a bespoke 'Sperm Morphology Assessment Standardisation Training Tool' to train novice morphologists using machine learning principles and consisted of two experiments. Experiment 1 assessed novice morphologists' (n = 22) accuracy across 2- category (normal; abnormal), 5- category (normal; head defect, midpiece defect, tail defect, cytoplasmic droplet), 8- category (normal; cytoplasmic droplet; midpiece defect; loose heads and abnormal tails; pyriform head; knobbed acrosomes; vacuoles and teratoids; swollen acrosomes), and 25- category (normal; all defects defined individually) classification systems, with untrained users achieving 81.0 ± 2.5%, 68 ± 3.59%, 64 ± 3.5%, and 53 ± 3.69%, respectively. A second cohort (n = 16) exposed to a visual aid and video significantly improved first-test accuracy (94.9 ± 0.66%, 92.9 ± 0.81%, 90 ± 0.91% and 82.7 ± 1.05, p < 0.001). Experiment 2 evaluated repeated training over four weeks, resulting in significant improvement in accuracy (82 ± 1.05% to 90 ± 1.38%, p < 0.001) and diagnostic speed (7.0 ± 0.4s to 4.9 ± 0.3s, p < 0.001). Final accuracy rates reached 98 ± 0.43%, 97 ± 0.58%, 96 ± 0.81%, and 90 ± 1.38% across classification systems 2-, 5-, 8- and 25-categories respectively. Significant differences in accuracy and variation were observed between the classification systems. This tool effectively standardised sperm morphology assessment. Future research could explore its application in other species, including in human andrology, given its accessibility and adaptability across classification systems.
精子形态学评估被认为是一项对男性生育能力至关重要但又存在变数的检测。这种变异性部分归因于形态学家缺乏标准化培训。本研究利用定制的“精子形态学评估标准化培训工具”，运用机器学习原理对新手形态学家进行培训，该研究包括两个实验。实验1评估了新手形态学家（n = 22）在2分类（正常；异常）、5分类（正常；头部缺陷、中段缺陷、尾部缺陷、细胞质滴）、8分类（正常；细胞质滴；中段缺陷；头部松散和尾部异常；梨形头；顶体有瘤；液泡和类畸形；顶体肿胀）和25分类（正常；单独定义的所有缺陷）分类系统中的准确率，未经培训的用户分别达到81.0±2.5%、68±3.59%、64±3.5%和53±3.69%。第二组（n = 16）在使用视觉辅助工具和视频后，首次测试的准确率显著提高（分别为94.9±0.66%、92.9±0.81%、90±0.91%和82.7±1.05%，p < 0.001）。实验2评估了为期四周的重复培训，结果准确率显著提高（从82±1.05%提高到90±1.38%，p < 0.001），诊断速度也显著提高（从7.0±0.4秒提高到4.9±0.3秒，p < 0.001）。最终，2分类、5分类、8分类和25分类系统的准确率分别达到98±0.43%、97±0.58%、96±0.81%和90±1.38%。各分类系统在准确率和变异性方面存在显著差异。该工具有效地实现了精子形态学评估的标准化。鉴于其在不同分类系统中的可及性和适应性，未来的研究可以探索其在包括人类男科学在内的其他物种中的应用。
