Comparative assessment and inter-platform workflow variability of bull sperm quality using benchtop and portable computer-assisted semen analysis systems

Computer-assisted semen analysis systems have improved the standardisation and objectivity of andrological diagnosis; however, workflow comparability between benchtop and portable platforms remains limited due to artefactual, methodological, optical, and operational differences among systems. This study assessed workflow comparability for bull sperm quality analysis. Thirty-two ejaculates from eleven Brahman bulls were collected by electroejaculation, using a standardised dilution and temperature control throughout sample processing. Sperm concentration was compared using Accuread®, Ai Station®, and iSperm®, whereas motility and kinematic variables were compared between Ai Station® and iSperm®, and sperm morphology was assessed using Ai Station® and Trumorph®. Differences between the workflows for each variable were evaluated using mixed-effects models, Bland-Altman analysis, and Passing-Bablok regression. Significant differences (P < 0.05) were observed among systems for sperm concentration, with Ai Station® yielding the highest estimates and iSperm® the lowest. Comparability for concentration was limited, with wide limits of agreement in the Bland-Altman plot, particularly at low-to-medium concentration ranges. For motility, Ai Station® produced higher total and progressive motility values than iSperm®. Likewise, curvilinear velocity, straight-line velocity, average path velocity, and linearity were higher in Ai Station®, whereas straightness was higher in iSperm®. In contrast, no significant differences were detected between Ai Station® and Trumorph® for normal sperm morphology; however, variability increased at low-to-intermediate normal morphology values, while divergence decreased when normal morphology exceeded 80%. Retrospective repeatability estimates showed moderate median CVs but wide upper ranges: 43–73% across concentration workflows and 90–95% across motility workflows; repeatability limits of agreement extended to approximately ±500 × 10⁶ spermatozoa/mL and more than ±30% points, respectively. Because no independent reference method was included, the results demonstrate divergence and non-interchangeability under the evaluated conditions, but not the accuracy or superiority of any platform.