Machine Learning-Based Analysis of Sperm Videos and Participant Data for Male Fertility Prediction

Abstract

Methods for automatic analysis of clinical data are usually targeted towards a specific modality and do not make use of all relevant data available. In the field of male human reproduction, clinical and biological data are not used to its fullest potential. Manual evaluation of a semen sample using a microscope is time-consuming and requires extensive training. Furthermore, the validity of manual semen analysis has been questioned due to limited reproducibility, and often high inter-personnel variation. The existing computer-aided sperm analyzer systems are not recommended for routine clinical use due to methodological challenges caused by the consistency of the semen sample. Thus, there is a need for an improved methodology. We use modern and classical machine learning techniques together with a dataset consisting of 85 videos of human semen samples and related participant data to automatically predict sperm motility. Used techniques include simple linear regression and more sophisticated methods using convolutional neural networks. Our results indicate that sperm motility prediction based on deep learning using sperm motility videos is rapid to perform and consistent. Adding participant data did not improve the algorithms performance. In conclusion, machine learning-based automatic analysis may become a valuable tool in male infertility investigation and research.

Deep Learning Based Evaluation of Spermatozoid Motility for Artificial Insemination

“…We also calculated the mean absolute error (MAE) of sperm vitality prediction. Following the suggestion presented in Hicks et al [ 19 ], the results with a mean MAE value below 11 were considered as significant improvements when compared to the ZeroR baseline, which assumes that the predicted values are equal to the average value computed over the dataset. Our results show that in all sperm samples we have achieved an average MAE value of less than 11 ( Figure 13 ), while the grand mean MAE was 2.92 (95% CI, 2.46–3.37).…”

“…For example, Nissen et al [ 18 ] compared common convolutional neural network (CNN) architectures for human sperm cell-segmentation and recognition in semen sample images, achieving 93.87% precision and 91.89% recall for the best-analyzed network architecture. Hicks et al [ 19 ] predicted the percentage of progressive, non-progressive, and immotile sperm heads from sperm images using ResNet-18 and ResNet-50 models and transfer learning. Movahed et al [ 20 ] used a combined CNN-kmeans-SVM approach to segment the exterior and interior parts of the sperm heads for quantitative morphological analysis of sperm heads.…”

Simplified sperm testing devices: a possible tool to overcome lack of accessibility and inconsistency in male factor infertility diagnosis. An opportunity for low- and middle- income countries.

“…Finally, to be fully applicable as a point-of-care screening tool, more research needs to be done on the dependence on an internet connection and the usability of these devices with washed sperm. Importantly, further equipping these devices with morphology assessment and improving their accuracy by use of diagnostic intelligence (AI) and putative sperm dysfunction testing (i.e DNA fragmentation assessment, Hyaluronan Binding Assay and Seminal Oxydative Stress) capabilities, offering a complete and timely quantitative and qualitative overview of a sperm sample, are ideal add-ons to further develop and ensure clinical mainstay of these innovative devices ( Agarwal et al, 2017 ; Dimitriadis et al, 2019 , Hicks et al, 2019 ).…”

Technological Advancements in Male Infertility Microsurgery

“…This is important because ART (including in-vitro fertilization and intracytoplasmic sperm injection), require optimal sperm selection for subsequent oocyte injection. While methods exist to aid in sperm selection, such as the swim-up or densitygradient centrifugation methods, there still remains significant subjectivity in sperm selection, which AI/ML may help optimize and make more objective [69,70]. Attempts have been made to automate semen analysis with computer-assisted semen analysis, but despite improved objectivity, it has been faced with decreased reproducibility, and, therefore, has not been adopted in many clinics for routine use [70,71].…”