Skip to main content
NCBI home page
Obstetrics & Gynecology Science logoLink to Obstetrics & Gynecology Science
. 2025 Nov 7;68(6):458–466. doi: 10.5468/ogs.25229

Innovations in assisted reproductive technology through the advances in artificial intelligence and image processing

Dayong Lee 1,2,✉
PMCID: PMC12645012  PMID: 41199550

Abstract

Advances in artificial intelligence (AI) and biomedical imaging have transformed reproductive medicine, offering new avenues for precision, efficiency, and objectivity in assisted reproductive technology (ART). Traditional embryo selection and ovarian stimulation monitoring rely mainly on subjective interpretation, which is often influenced by inter- and intra-observer variability. In contrast, AI-enhanced models have demonstrated consistent performance, reduced human-dependent discrepancies, and improved reproducibility of clinical decisions. Among emerging technologies, fluorescence lifetime imaging microscopy enables real-time, label-free metabolic assessment of gametes and embryos by quantifying the intrinsic fluorescence lifetimes of nicotinamide adenine dinucleotide phosphate and flavine adenine dinucleotide. These metabolic signatures correlate with developmental competence, providing a non-invasive tool to evaluate embryo quality beyond the morphological criteria. Recent innovations have extended AI and imaging technologies to self-operated reproductive health monitoring. Studies support AI-powered self-assessment of ovarian follicles using smartphone-compatible ultrasound devices and automated follicle segmentation. This development has potential for improving ovarian stimulation tracking, patient engagement, and personalizing treatment protocols in clinical and low-resource settings. The integration of AI, advanced image processing, and metabolic imaging is a promising frontier in reproductive medicine. These tools enhance the precision of embryo and follicle evaluation, while establishing foundations multimodal platforms that combine clinical, morphological, and metabolic data to optimize ART outcomes.

Keywords: Artificial intelligence, Reproductive medicine, Embryo, Time-lapse imaging, In vitro fertilization

Introduction

Reproductive medicine has evolved remarkably over recent decades, largely driven by the integration of novel biomedical technologies. From in vitro fertilization (IVF) to vitrification techniques, assisted reproductive technologies (ARTs) have transformed infertility treatments into highly personalized and data-intensive domains. Despite these advances, clinicians face challenges related to implantation rate plateaus, subjective nature of embryo selection, and limited predictive value of current morphological criteria [1]. In response to these limitations, artificial intelligence (AI) and advanced image-processing technologies have emerged as promising tools for improving decision-making and clinical outcomes in reproductive medicine. Similarly, in obstetrics and gynecologic oncology, research explores using AI to predict pregnancy complications, detect gynecologic cancers at an early stage, and forecast recurrence, which are relevant in the field of reproductive endocrinology [2–5]. AI, particularly deep learning, offers powerful data-driven solutions for detecting subtle patterns in complex biological datasets. These models have been deployed to automate embryo grading, predict clinical pregnancy outcomes, and optimize ovarian stimulation protocols [6–8]. Advancements in high-resolution imaging and real-time metabolic assessment have opened new avenues for non-invasive embryo viability analysis [9].

As these technologies converge, the field of reproductive medicine transitions from morphology-based judgment to an integrative, objective, and quantifiable model of patient and embryo assessment. This review aimed to explore recent innovations in AI and image processing in reproductive medicine, focusing on their applications in embryo selection, gamete evaluation, and predictive modeling for IVF outcomes. We discuss the emerging role of fluorescence lifetime imaging microscopy (FLIM) in non-invasive metabolic profiling and its potential for clinical translation [10]. By highlighting the synergies among these technologies, we highlight their impact on enhancing precision, efficiency, and success in fertility care.

AI in reproductive medicine

AI, particularly deep learning, has emerged as a transformative tool in reproductive medicine. By leveraging large volumes of clinical, biochemical, physiological, and imaging data, AI systems can identify complex patterns beyond human perception, offering new strategies to optimize patient care and improve success rates. This section highlights key applications of AI in embryo selection, predictive modeling, and gamete evaluation.

1. AI in embryo selection and grading

Traditional embryo selection relies heavily on subjective morphological assessments, leading to inter-observer variability and suboptimal outcomes. Time-lapse imaging introduces more objective morphokinetic parameters; however, interpretation remains labor-intensive and operator-dependent. Recently, AI-driven platforms such as convolutional neural networks (CNNs) have automated embryo assessment using thousands of time-lapse images. Recent commercially available embryo assessment systems use deep learning to predict the implantation potential using a vast database of annotated embryo images. In a large retrospective study, Tran et al. [8] demonstrated that AI outperformed embryologists in predicting clinical pregnancies after embryo transfer. Similarly, Bormann et al. [7] reported that a deep learning model trained on 3,469 video records of embryos collected from 543 patients achieved predictive performance comparable to that of trained embryologists. These AI systems improve consistency in grading and facilitate single embryo transfer by identifying viable embryos, reducing multiple gestation risks.

2. Predictive modeling for clinical outcomes

AI has been utilized to predict complex clinical outcomes such as ovarian response, fertilization rates, and pregnancy probability in IVF. By integrating parameters such as age, anti-Müllerian hormone levels, antral follicle counts (AFCs), and stimulation protocols, machine-learning algorithms can generate personalized predictions. Khosravi et al. [6] used a multimodal AI model that combined imaging features with clinical data to enhance successful implantation prediction. These models can aid clinicians in counseling patients, tailoring stimulation regimens, and determining the optimal day and number of embryos to transfer. Additionally, explainable AI approaches have been developed to increase clinical trust by visualizing decision pathways and highlighting their predictive features.

3. AI in sperm assessment

Sperm quality crucial for male fertility assessments. Conventional semen analysis, guided by World Health Organization criteria, involves subjective evaluation of sperm concentration, motility, morphology, and vitality. AI-based image analysis has emerged as a powerful tool for the automation and enhancement of sperm quality.

AI applications for sperm evaluation focus primarily on morphological classification and motility tracking. Deep-learning models, particularly CNNs, show high accuracy in classifying sperm head shapes, tail defects, and midpiece abnormalities based on microscopic images [11]. Recent studies have developed automated image analysis pipelines using machine learning to accurately classify sperm motility patterns, achieving comparable performance to that of manual annotation [12–14].

Recent research has explored the integration of label-free imaging techniques, such as holographic microscopy and quantitative phase imaging, with AI to extract intrinsic biophysical features from live sperm. These approaches eliminate the need for staining, allowing real-time, non-destructive assessment of sperm integrity and function. In a study by Rivenson et al. [15], a deep learning model was trained to reconstruct high-resolution sperm images from lens-free holograms and distinguish between normal and abnormal morphologies with high precision.

4. AI in oocyte assessment

The accurate assessment of human oocyte quality is a critical for IVF success. Traditionally, oocyte quality has been evaluated subjectively by embryologists based on morphological parameters, such as cytoplasmic texture, zona pellucida thickness, polar body appearance, and the presence of vacuoles or granularity. However, this approach has limited reproducibility, significant inter-observer variability, and low predictive power for downstream outcomes such as fertilization or embryo development.

To overcome these limitations, recent studies have explored the application of AI-based image analysis to objectively evaluate oocyte quality. These models can process large datasets of bright-field or polarized light microscopy images to detect features related to oocyte quality [16]. These quantitative features were integrated into a predictive algorithm that outperformed embryologist-based grading.

Advances in image processing technologies

Image processing has played a pivotal role in reproductive medicine, particularly in assessing gametes, embryos, and endometrial receptivity. With digital microscopy, time-lapse imaging, and automated analysis, recent advances enable more objective, reproducible, and dynamic insights into early developmental biology. These improvements increase diagnostic accuracy and enhance clinical decision-making, when integrated with AI-based tools.

1. Time-lapse imaging and morphokinetics

Time-lapse incubation systems have revolutionized embryo assessment by enabling continuous monitoring without removing the embryos from stable culture conditions. These platforms capture thousands of images across multiple focal planes throughout preimplantation development, allowing clinicians to extract morphokinetic parameters (e.g., time to 2-cell stage, compaction onset, and blastocyst formation).

Several studies have shown that certain morphokinetic patterns associate with higher implantation potential. Basile and Meseguer [17] identified specific cleavage intervals (t5-t3 and t3-t2) that correlate with clinical pregnancy outcomes, suggesting temporal dynamics can serve as objective biomarkers. However, interpretation remains complex and subject to inter-laboratory variation, leading to the integration of automated image analysis with deep learning models to standardize the evaluations.

2. High-resolution and label-free imaging modalities

Beyond time-lapse systems, several high-resolution imaging technologies enable non-invasive assessment of embryos and gametes. These include optical coherence tomography, which enables the three-dimensional visualization of internal embryonic structures without staining [18]. Additionally, quantitative phase imaging captures refractive index variations to assess cellular morphology and dynamic behavior [19]. These label-free techniques avoid phototoxicity and do not compromise the embryo viability. They offer detailed insights into subcellular structures difficult to detect using conventional bright-field microscopy.

3. Automated feature extraction and standardization

A key bottleneck in image-based embryo assessment is the subjective feature selection. Recent algorithms automatically extract features such as blastocyst area, zona thickness, cellular symmetry, cytoplasmic granularity, and vacuolization using image segmentation techniques [2,20]. These quantitative descriptors allow consistent embryo annotation and ranking, particularly in high-volume IVF clinics. Moreover, integration with AI enables real-time feedback to embryologists and can alert clinicians when embryos deviate from normal developmental trajectories, potentially preventing poor-quality embryo transfers.

4. AI-enabled monitoring and self-monitoring of ovarian follicles

Manual follicle measurement is time-consuming and prone to high inter- and intra-observer variabilities. To overcome these limitations, recent studies have explored deep learning algorithms to quantify follicular areas from ultrasound images as novel biomarkers for automated follicular monitoring [21]. Furthermore, advances in AI-based image analysis have enabled three-dimensional ultrasound to estimate follicle volume as a potential indicator of oocyte maturity [22]. During ovarian stimulation for ART, both the AFC and follicular volume are routinely tracked. However, volumetric analysis of the ovary and follicles remains operator dependent and prone to errors. To address this issue, ongoing research focuses on developing deep-learning-based methods for automatic and simultaneous segmentation of ovaries and follicles in three-dimensional transvaginal ultrasound images [23]. Moreover, AI-driven systems have been proposed to automatically identify and monitor follicle growth trajectories using serial ultrasound images acquired during ovarian stimulation [24].

Recent efforts have been made to assess the ovarian reserve through self-administered virtual transvaginal ultrasonography at home using portable transvaginal ultrasound probes connected to smartphones, remotely guided by certified ultrasound technologists. Studies show this approach yields results noninferior to those of traditional in-clinic transvaginal ultrasonography [25].

These systems may enable patients to remotely track follicular development with high accuracy and speed using AI-assisted tools during ovarian stimulation, potentially transforming clinical workflow and patient experience in ART.

Application of FLIM in reproductive medicine

The success of ART depends on accurate selection of developmentally competent gametes and embryos. Although they are widely used, traditional morphological assessments lack predictive precision. FLIM, a label-free quantitative metabolic imaging modality, has emerged as a promising tool for assessing the health and viability of oocytes, embryos, and supporting cells.

1. Principles of FLIM in reproductive cells

FLIM measures fluorescence decay time emitted by intrinsic cellular cofactors, nicotinamide adenine dinucleotide phosphate (NAD(P)H) and flavin adenine dinucleotide (FAD) [10]. These lifetimes reflect the metabolic state of the cells, particularly the balance between glycolysis and oxidative phosphorylation. Unlike fluorescence intensity, lifetime measurements are independent of fluorophore concentration or optical path variability, providing reproducible and quantitative insights into cellular functions. A recent advancement in FLIM analysis is the phasor approach, which transforms fluorescence decay data into a graphical representation without requiring exponential fitting, enhancing speed and accuracy for clinical embryo evaluation.

2. Applications in cumulus cells, oocytes, and embryos

Studies have been published using FLIM to measure their metabolic status of mouse oocytes or embryos [26–28]. Research findings have emerged on FLIM applications in human reproductive cells based on these studies (Table 1). As reported by Venturas et al. [29], FLIM profiling of cumulus cells (CCs) can noninvasively predict oocyte maturity. Lifetime parameters, such as the bound NAD(P)H fraction, vary between immature and mature oocyte-associated CCs and can detect age-related metabolic shifts in human oocytes, including increased oxidative stress and reduced mitochondrial efficiency. These changes are linked to poor developmental outcomes and can be detected without compromising oocyte integrity. Venturas et al. [30] showed that these metabolic profiles are correlate with pregnancy and live birth outcomes. Moreover, FLIM has emerged as an effective tool for assessing the quality of pre-implantation embryos. Shah et al. [31] and Venturas et al. [32] demonstrated that FLIM-derived redox ratios and bound NAD(P)H fractions are associated with embryo metabolic status and euploidy. This method enables rapid, label-free assessment of embryo metabolic activity, providing quantitative markers of embryo health. Their findings suggest that FLIM analysis can classify blastocysts with high predictive accuracy, even in vitrified-warmed embryos. Further studies on implementing FLIM for embryo selection in clinical IVF are warranted.

Table 1.

Fluorescence lifetime imaging microscopy studies in reproductive biology

Study Cell type/source Key findings Clinical relevance
Sanchez et al. [26] (2018) Oocytes/mouse FLIM revealed metabolic differences correlating with developmental potential Non-invasive tool to assess mitochondrial function in oocytes
Ma et al. [27] (2019) Embryos of different developmental stage/mouse Phasor-FLIM predicted embryo viability with high accuracy Enhancing embryo selection processes
Seidler et al. [28] (2020) Oocytes and embryos/mouse Metabolic response to transient hypoxia could be detected using FLIM Accessing metabolic plasticity as preimplantation embryos develop
Venturas et al. [29] (2021) Cumulus cells/human FLIM parameters correlated with age and AMH; significant differences between cumulus cells from mature vs. immature oocytes Potential non-invasive biomarker of cumulus cells for oocyte maturity and quality
Venturas et al. [32] (2022) Blastocysts/human FLIM detected metabolic heterogeneity between inner cell mass and trophectoderm and across developmental stages Potential of FLIM for assessing embryo quality and viability non-invasively
Shah et al. [31] (2022) Blastocysts/human FLIM detected metabolic differences between euploid and aneuploid blastocysts Non-invasive clinical tool to identify the euploid embryos
Vishnyakova et al. [33] (2023) Sperm/human Different sperm motility was related to the type of their energy metabolism assessed using FLIM Non-invasive and gentle procedure to segregate the best spermatozoa beyond motility and morphology
Venturas et al. [30] (2024) Cumulus cells/human FAD+FLIM parameters of cumulus cells were significantly associated with morphological rank of blastocysts and embryos resulting in a clinical pregnancy Detecting oocytes leading to embryos that result in a clinical pregnancy and a live birth
Open in a new tab

FLIM, fluorescence lifetime imaging microscopy; AMH, anti-Müllerian hormone; FAD, flavin adenine dinucleotide.

3. Applications in sperm analysis

FAD serves as a key cofactor for the flavoenzymes in redox reactions and is crucial for sperm metabolism. FAD is predominantly localized in the mitochondria and essential for maintaining normal sperm function and motility. Given its metabolic importance, assessing FAD levels and its microenvironment via FLIM is a promising non-invasive biomarker for evaluating sperm quality.

Vishnyakova et al. [33] applied FLIM to human spermatozoa and reported significant differences in NAD(P)H fluorescence lifetimes between high- and low-motility sperm populations, highlighting the utility of FLIM for the functional assessment of male gametes. This approach could complement traditional morphology-based evaluation methods and when integrated with AI-assisted image interpretation, enable automated sperm selection in ART.

4. Noninvasive metabolic profiling in reproduction using FLIM

FLIM has progressed from an experimental tool to a candidate clinical tool for ART. With advances such as phasor-based FLIM analysis, metabolic imaging enables a rapid, noninvasive, and spatially resolved assessment of embryo viability and quality. The integration of AI has further enhanced the clinical applicability of FLIM by enabling the automated and objective interpretation of complex metabolic image data. Studies have demonstrated correlations between FLIM-derived metabolic signatures and outcomes, such as ploidy, pregnancy, and live birth. When integrated with AI-driven workflows, FLIM can revolutionize oocyte, sperm, and embryo selection, offering a safe and effective alternative to conventional methods.

Challenges and limitations

Despite the advances of AI and imaging techniques in reproductive medicine, several significant challenges must be addressed before these tools can be integrated into routine clinical practice. These challenges span technical, regulatory, ethical, and clinical domains, and their resolution is essential to ensure the safety, equity, and reliability of ART.

1. Lack of standardization and validation

A major limitation of AI-driven embryo or gamete assessment is the lack of standardized datasets and uniform imaging protocols. AI models are trained using data from a few clinics or laboratories, limiting their generalizability across different populations and imaging systems [34]. Morphokinetic variables derived from time-lapse imaging may vary with culture conditions, incubator type, or media, complicating cross-platform comparisons and weakens reproducibility. Similarly, FLIM remains largely experimental in human ART due to variability in equipment calibration, fluorophore signal interpretation, and lifetime quantification standards. Without robust multicenter validation and clinical trials, regulatory approval and widespread adoption remain important goals.

2. Interpretability and clinical trust

Many deep-learning models, particularly CNNs, operate as “black boxes”, producing predictions without offering clear explanations for their decisions. This lack of interpretability can hinder clinician confidence and may be problematic for ethical and legal accountability [35]. Recent efforts in explainable AI have aimed to make these systems more transparent by highlighting the image regions or features that drive the predictions.

3. Ethical and privacy concerns

The use of AI in reproductive medicine raises privacy and consent concerns, especially sensitive data such as patient demographics, embryo images, and genomic information are collected and processed. Data governance frameworks that ensure deidentification, secure storage, and controlled access are not yet universally established [36]. Moreover, a concern exists that automated embryo selection systems may reinforce biases in training datasets, disadvantaging certain populations, or prioritizing embryos based on non-clinical features such as morphology that are weakly associated with live birth rates [37].

4. Cost and accessibility

Advanced imaging platforms (e.g., time-lapse incubators) and AI software solutions require substantial financial investments, specialized training, and infrastructure. This limits accessibility in low-resource settings and may exacerbate global disparities in fertility care. In addition, the maintenance of these technologies and output interpretation typically require a multidisciplinary team involving embryologists, data scientists, and engineers.

5. Regulatory and legal barriers

To date, few AI tools for embryo selection have received formal regulatory approval from agencies such as the U.S. Food and Drug Administration or European Conformité Européenne certification [38]. As AI models evolve rapidly, regulatory frameworks often lag, creating uncertainty regarding liabilities, validation requirements, and clinical deployment. Establishing consistent guidelines for algorithm approval, model retraining, and performance auditing is critical for clinical translation.

Future directions

The convergence of AI, advanced imaging, and metabolic sensing technologies has heralded a new era precision reproductive care. As data volume and computational power grow, several promising directions are emerging.

1. Integration of multimodal data

A key avenues lies in the integrating diverse data types-clinical profiles, morphokinetic patterns, metabolic signatures from FLIM, hormonal responses, and genetic information-into unified AI models. These models can deliver personalized embryo selection, stimulation protocols, and treatment plans tailored to each patient’s unique biology. This approach may enable real-time decision support in IVF laboratories, offering embryologists predictive probabilities for embryo implantation potential or recommending transfer strategies based on dynamic changes in embryo development and metabolic activity.

2. Clinical translation of FLIM and non-invasive screening

FLIM, currently limited to research settings, is poised for clinical application as instrumentation becomes more precise, standardized, and user-friendly. When paired with conventional imaging, FLIM can provide a real-time, label-free embryo viability assessment without invasive procedures such as blastomere biopsy [39]. Additionally, FLIM can evaluate oocyte aging, endometrial receptivity, and sperm oxidative stress, expanding its utility across the reproductive process.

3. AI in low-resource settings

As cloud computing and mobile imaging technologies become affordable, simplified AI platforms may be adapted for use in low- and middle-income countries. Portable AI-enabled diagnostic tools can democratize access to infertility evaluation, basic embryology, or ovulation monitoring, helping to address global disparities in fertility care [40].

4. Regulatory and ethical maturation

To achieve long-term success, AI and imaging systems must evolve within clear ethical and legal frameworks. These include developing transparent algorithms, ensuring data privacy, and establishing international standards for model validation and clinical performance. Collaboration among clinicians, data scientists, and regulatory bodies is essential to balance innovation with safety and equity.

Conclusion

AI and image-based technologies are reshaping the landscape of reproductive medicine. From improving embryo selection and personalizing treatment to enabling non-invasive metabolic assessment via FLIM, these innovations enhance clinical outcomes, reduce procedural risks, and enable precision fertility care (Fig. 1). In the future, technical limitations, ethical concerns, and infrastructure barriers should be addressed through interdisciplinary collaboration and validation studies. By fostering responsible innovation, integrating AI and advanced imaging can create a new paradigm in reproductive health that is intelligent, individualized, and inclusive.

Fig. 1.

Fig. 1

Open in a new tab

AI-driven precision medicine in ART. FLIM, fluorescence lifetime imaging microscopy; AI, artificial intelligence; ART, assisted reproductive technology; LLM, large language model.

Footnotes

Conflict of interest

No potential conflict of interest relevant to this article was reported.

Ethical approval

Not applicable.

Patient consent

Not applicable.

Funding information

None.

Acknowledgement

This review article was prepared by the Korean Society of Ultrasound in Obstetrics and Gynecology as an affiliated society of the Korean Society of Obstetrics and Gynecology.

References

  • 1.Siristatidis C, Pouliakis A, Chrelias C, Kassanos D. Artificial intelligence in IVF: a need. Syst Biol Reprod Med. 2011;57:179–85. doi: 10.3109/19396368.2011.558607. [DOI] [PubMed] [Google Scholar]
  • 2.Ahn KH, Lee KS. Artificial intelligence in obstetrics. Obstet Gynecol Sci. 2022;65:113–24. doi: 10.5468/ogs.21234. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Lee Y, Kim SY. Potential applications of ChatGPT in obstetrics and gynecology in Korea: a review article. Obstet Gynecol Sci. 2024;67:153–9. doi: 10.5468/ogs.23231. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Abrar SS, Isa SAM, Hairon SM, Ismail MP, Kadir MNBNA. Recent advances in applications of machine learning in cervical cancer research: a focus on prediction models. Obstet Gynecol Sci. 2025;68:247–59. doi: 10.5468/ogs.25041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Akazawa M, Hashimoto K, Noda K, Yoshida K. The application of machine learning for predicting recurrence in patients with early-stage endometrial cancer: a pilot study. Obstet Gynecol Sci. 2021;64:266–73. doi: 10.5468/ogs.20248. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Khosravi P, Kazemi E, Zhan Q, Malmsten JE, Toschi M, Zisimopoulos P, et al. Deep learning enables robust assessment and selection of human blastocysts after in vitro fertilization. NPJ Digit Med. 2019;2:21. doi: 10.1038/s41746-019-0096-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Bormann CL, Kanakasabapathy MK, Thirumalaraju P, Gupta R, Pooniwala R, Kandula H, et al. Performance of a deep learning based neural network in the selection of human blastocysts for implantation. Elife. 2020;9:e55301. doi: 10.7554/eLife.55301. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Tran D, Cooke S, Illingworth PJ, Gardner DK. Deep learning as a predictive tool for fetal heart pregnancy following time-lapse incubation and blastocyst transfer. Hum Reprod. 2019;34:1011–18. doi: 10.1093/humrep/dez064. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.VerMilyea M, Hall JMM, Diakiw SM, Johnston A, Nguyen T, Perugini D, et al. Development of an artificial intelligence-based assessment model for prediction of embryo viability using static images captured by optical light microscopy during IVF. Hum Reprod. 2020;35:770–84. doi: 10.1093/humrep/deaa013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Venturas M, Yang X, Sakkas D, Needleman D. Noninvasive metabolic profiling of cumulus cells, oocytes, and embryos via fluorescence lifetime imaging microscopy: a mini-review. Hum Reprod. 2023;38:799–810. doi: 10.1093/humrep/dead063. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.You JB, McCallum C, Wang Y, Riordon J, Nosrati R, Sinton D. Machine learning for sperm selection. Nat Rev Urol. 2021;18:387–403. doi: 10.1038/s41585-021-00465-1. [DOI] [PubMed] [Google Scholar]
  • 12.Ottl S, Amiriparian S, Gerczuk M, Schuller BW. motilitAI: a machine learning framework for automatic prediction of human sperm motility. iScience. 2022;25:104644. doi: 10.1016/j.isci.2022.104644. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Hicks SA, Andersen JM, Witczak O, Thambawita V, Halvorsen P, Hammer HL, et al. Machine learning-based analysis of sperm videos and participant data for male fertility prediction. Sci Rep. 2019;9:16770. doi: 10.1038/s41598-019-53217-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Javadi S, Mirroshandel SA. A novel deep learning method for automatic assessment of human sperm images. Comput Biol Med. 2019;109:182–94. doi: 10.1016/j.compbiomed.2019.04.030. [DOI] [PubMed] [Google Scholar]
  • 15.Rivenson Y, Liu T, Wei Z, Zhang Y, de Haan K, Ozcan A. PhaseStain: the digital staining of label-free quantitative phase microscopy images using deep learning. Light Sci Appl. 2019;8:23. doi: 10.1038/s41377-019-0129-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Amini Mahabadi J, Enderami SE, Nikzad H, Hassani Bafrani H. The use of machine learning for human sperm and oocyte selection and success rate in IVF methods. Andrologia. 2024;2024:8165541. [Google Scholar]
  • 17.Basile N, Meseguer M. Time-lapse technology: evaluation of embryo quality and new markers for embryo selection. Expert Rev Obstet Gynecol. 2012;7:175–90. [Google Scholar]
  • 18.Morawiec S, Ajduk A, Stremplewski P, Kennedy BF, Szkulmowski M. Full-field optical coherence microscopy enables high-resolution label-free imaging of the dynamics of live mouse oocytes and early embryos. Commun Biol. 2024;7:1057. doi: 10.1038/s42003-024-06745-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Park J, Bai B, Ryu D, Liu T, Lee C, Luo Y, et al. Artificial intelligence-enabled quantitative phase imaging methods for life sciences. Nat Methods. 2023;20:1645–60. doi: 10.1038/s41592-023-02041-4. [DOI] [PubMed] [Google Scholar]
  • 20.Diakiw SM, Hall JMM, VerMilyea MD, Amin J, Aizpurua J, Giardini L, et al. Development of an artificial intelligence model for predicting the likelihood of human embryo euploidy based on blastocyst images from multiple imaging systems during IVF. Hum Reprod. 2022;37:1746–59. doi: 10.1093/humrep/deac131. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Liang X, Fang J, Li H, Yang X, Ni D, Zeng F, et al. CR-unet-based ultrasonic follicle monitoring to reduce diameter variability and generate area automatically as a novel biomarker for follicular maturity. Ultrasound Med Biol. 2020;46:3125–34. doi: 10.1016/j.ultrasmedbio.2020.07.020. [DOI] [PubMed] [Google Scholar]
  • 22.Liang X, Liang J, Zeng F, Lin Y, Li Y, Cai K, et al. Evaluation of oocyte maturity using artificial intelligence quantification of follicle volume biomarker by three-dimensional ultrasound. Reprod Biomed Online. 2022;45:1197–206. doi: 10.1016/j.rbmo.2022.07.012. [DOI] [PubMed] [Google Scholar]
  • 23.Mathur P, Kakwani K, Diplav, Kudavelly S, Ga R. Deep learning based quantification of ovary and follicles using 3d transvaginal ultrasound in assisted reproduction. Annu Int Conf IEEE Eng Med Biol Soc. 2020;2020:2109–12. doi: 10.1109/EMBC44109.2020.9176703. [DOI] [PubMed] [Google Scholar]
  • 24.Srivastava D, Gupta S, Kudavelly S, KVS, Ga R. Unsupervised deep learning based longitudinal follicular growth tracking during IVF cycle using 3D transvaginal ultrasound in assisted reproduction. Annu Int Conf IEEE Eng Med Biol Soc. 2021;2021:3209–12. doi: 10.1109/EMBC46164.2021.9630495. [DOI] [PubMed] [Google Scholar]
  • 25.Chung EH, Petishnok LC, Conyers JM, Schimer DA, Vitek WS, Harris AL, et al. Virtual compared with in-clinic transvaginal ultrasonography for ovarian reserve assessment. Obstet Gynecol. 2022;139:561–70. doi: 10.1097/AOG.0000000000004698. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Sanchez T, Wang T, Pedro MV, Zhang M, Esencan E, Sakkas D, et al. Metabolic imaging with the use of fluorescence lifetime imaging microscopy (FLIM) accurately detects mitochondrial dysfunction in mouse oocytes. Fertil Steril. 2018;110:1387–97. doi: 10.1016/j.fertnstert.2018.07.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Ma N, Mochel NR, Pham PD, Yoo TY, Cho KWY, Digman MA. Label-free assessment of pre-implantation embryo quality by the fluorescence lifetime imaging microscopy (FLIM)-phasor approach. Sci Rep. 2019;9:13206. doi: 10.1038/s41598-019-48107-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Seidler EA, Sanchez T, Venturas M, Sakkas D, Needleman DJ. Non-invasive imaging of mouse embryo metabolism in response to induced hypoxia. J Assist Reprod Genet. 2020;37:1797–805. doi: 10.1007/s10815-020-01872-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Venturas M, Yang X, Kumar K, Wells D, Racowsky C, Needleman DJ. Metabolic imaging of human cumulus cells reveals associations among metabolic profiles of cumulus cells, patient clinical factors, and oocyte maturity. Fertil Steril. 2021;116:1651–62. doi: 10.1016/j.fertnstert.2021.07.1204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Venturas M, Racowsky C, Needleman DJ. Metabolic imaging of human cumulus cells reveals associations with pregnancy and live birth. Hum Reprod. 2024;39:1176–85. doi: 10.1093/humrep/deae087. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Shah JS, Venturas M, Sanchez TH, Penzias AS, Needleman DJ, Sakkas D. Fluorescence lifetime imaging microscopy (FLIM) detects differences in metabolic signatures between euploid and aneuploid human blastocysts. Hum Reprod. 2022;37:400–10. doi: 10.1093/humrep/deac016. [DOI] [PubMed] [Google Scholar]
  • 32.Venturas M, Shah JS, Yang X, Sanchez TH, Conway W, Sakkas D, et al. Metabolic state of human blastocysts measured by fluorescence lifetime imaging microscopy. Hum Reprod. 2022;37:411–27. doi: 10.1093/humrep/deab283. [DOI] [PubMed] [Google Scholar]
  • 33.Vishnyakova P, Nikonova E, Jumaniyazova E, Solovyev I, Kirillova A, Farmakovskaya M, et al. Fluorescence lifetime imaging microscopy as an instrument for human sperm assessment. Biochem Biophys Res Commun. 2023;645:10–6. doi: 10.1016/j.bbrc.2023.01.016. [DOI] [PubMed] [Google Scholar]
  • 34.Wang R, Pan W, Jin L, Li Y, Geng Y, Gao C, et al. Artificial intelligence in reproductive medicine. Reproduction. 2019;158:R139–54. doi: 10.1530/REP-18-0523. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Holzinger A, Langs G, Denk H, Zatloukal K, Müller H. Causability and explainability of artificial intelligence in medicine. Wiley Interdiscip Rev Data Min Knowl Discov. 2019;9:e1312. doi: 10.1002/widm.1312. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Cohen IG, Amarasingham R, Shah A, Xie B, Lo B. The legal and ethical concerns that arise from using complex predictive analytics in health care. Health Aff (Millwood) 2014;33:1139–47. doi: 10.1377/hlthaff.2014.0048. [DOI] [PubMed] [Google Scholar]
  • 37.Rolfes V, Bittner U, Gerhards H, Krüssel JS, Fehm T, Ranisch R, et al. Artificial intelligence in reproductive medicine - an ethical perspective. Geburtshilfe Frauenheilkd. 2023;83:106–15. doi: 10.1055/a-1866-2792. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Fraser H, Coiera E, Wong D. Safety of patient-facing digital symptom checkers. Lancet. 2018;392:2263–4. doi: 10.1016/S0140-6736(18)32819-8. [DOI] [PubMed] [Google Scholar]
  • 39.König K. Review: clinical in vivo multiphoton FLIM tomography. Methods Appl Fluoresc. 2020;8:034002. doi: 10.1088/2050-6120/ab8808. [DOI] [PubMed] [Google Scholar]
  • 40.Gbagbo FY, Ameyaw EK, Yaya S. Artificial intelligence and sexual reproductive health and rights: a technological leap towards achieving sustainable development goal target 3.7. Reprod Health. 2024;21:196. doi: 10.1186/s12978-024-01924-9. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Obstetrics & Gynecology Science are provided here courtesy of Korean Society of Obstetrics and Gynecology

RESOURCES