AI Analysis of 3D Ultrasound for IVF Outcomes

(AI in ART Trial)

Overseen By Rodriq Stubbs, NP

Age: 18+

Sex: Any

Trial Phase: Academic

Sponsor: Weill Medical College of Cornell University

No Placebo Group All trial participants will receive the active study treatment (no placebo)

Compensation: Varies

Number of Visits: 1

Duration: 6-8 weeks

4000 Participants Needed

Trial Summary

What is the purpose of this trial?

The use of machine learning techniques using an artificial intelligence tool is proposed to analyze clinical data to predict best possible IVF/ART outcomes. This tool has been utilized to accurately predict embryo quality here at Cornell. Utilizing this tool to assess objective clinical findings and predict outcomes of assisted reproductive techniques is sought, with the ultimate goal of an automated tool to reduce implicit physician bias. Within this goal, using this tool to objectively and accurately assess baseline ovarian reserve at the start of an ART cycle is proposed, using 3D sonography to image the ovary and artificial intelligence tool to objectively identify baseline antral follicle counts.

Will I have to stop taking my current medications?

The trial information does not specify whether you need to stop taking your current medications.

What data supports the effectiveness of the treatment AI to analyze 3D ultrasound for IVF outcomes?

AI has shown promise in improving various aspects of IVF, such as optimizing workflow during ovarian stimulation, predicting embryo quality, and enhancing embryo selection processes, which can lead to higher pregnancy rates. AI's ability to analyze large amounts of data allows for personalized IVF treatments, potentially improving outcomes. ^{1 2 3 4 5}

Is AI analysis of 3D ultrasound for IVF outcomes safe for humans?

The research articles do not provide specific safety data for AI analysis of 3D ultrasound in humans, but AI methods have been implemented in various aspects of IVF without reported safety concerns. ^{1 2 3 5 6}

How does the AI analysis of 3D ultrasound for IVF outcomes differ from other treatments?

This treatment is unique because it uses artificial intelligence to analyze 3D ultrasound images, potentially providing more personalized and accurate predictions of IVF outcomes compared to traditional methods. Unlike standard treatments, which rely heavily on human interpretation, this approach leverages AI's ability to process large amounts of data to optimize IVF procedures and improve success rates. ^{1 2 3 5 7}

Research Team

Nikica Zaninovic, PHD

Principal Investigator

Weill Medical College of Cornell University

Eligibility Criteria

This trial is for patients undergoing ovarian stimulation treatments, including those aiming for fresh embryo transfer and cryopreservation of oocytes or embryos. Healthy male partners of female subjects who consent can also participate.

Inclusion Criteria

I am a healthy male partner agreeing to participate in the study.

All patients undergoing ovarian stimulation (including OI and IVF cycles)

I am undergoing treatment for embryo transfer and egg or embryo freezing.

Exclusion Criteria

Not applicable.

Timeline

Screening

Participants are screened for eligibility to participate in the trial

2-4 weeks

Baseline Evaluation

3D transvaginal ultrasound of ovaries at the beginning of an ART cycle to assess baseline ovarian reserve

1 day

1 visit (in-person)

Treatment

Participants undergo ART cycle with AI tool assessing antral follicle count and monitoring of IVF outcomes

6-8 weeks

Follow-up

Participants are monitored for pregnancy outcomes and other ART-related results

up to 9 months

Treatment Details

Interventions

• AI to analyze 3 D ultrasound

Trial Overview The study tests an AI tool that analyzes 3D ultrasound images to predict the best IVF/ART outcomes. It aims to create an automated system that reduces physician bias by objectively assessing baseline ovarian reserve at the start of ART cycles.

Participant Groups

Group I: 3D Ultrasound with AI Experimental Treatment 1 Intervention

AI tool to assess antral follicle count using 3 D Ultrasound

Who Is Running the Clinical Trial?

Weill Medical College of Cornell University

Lead Sponsor

Trials

1,103

Recruited

1,157,000+

Findings from Research

Recent advancements in IVF have incorporated artificial intelligence (AI) to analyze large datasets, aiming to personalize treatments and improve outcomes, particularly in embryo selection and stimulation protocols.

Despite these technological advancements, the embryo transfer process remains reliant on human skill, highlighting the importance of manual techniques and decision-making in achieving successful IVF results.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

In Contemporary Reproductive Medicine Human Beings are Not Yet Dispensable. Allahbadia, GN., Allahbadia, SG., Gupta, A.[2023]

A multi-modal fusion model that combines ultrasound-based deep learning radiomics with clinical parameters significantly improves the prediction of clinical pregnancy outcomes after frozen embryo transfer (FET) in a study of 240 women, outperforming traditional methods that use either imaging or clinical data alone.

The model achieved a high area under the curve (0.825) and demonstrated strong sensitivity (96.2%) and specificity (58.3%), indicating its potential as a reliable and non-invasive tool for personalized evaluation of endometrial receptivity.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

An ultrasound-based deep learning radiomic model combined with clinical data to predict clinical pregnancy after frozen embryo transfer: a pilot cohort study. Liang, X., He, J., He, L., et al.[2023]

A refined endometrial wave classification system was developed from ultrasound images of 24 IVF patients, introducing two new wave types: recoiling CF waves and standing waves.

The study demonstrated strong agreement between different observers when classifying endometrial wave types, indicating that the new system can reliably describe complex wave patterns during IVF cycles.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Complex endometrial wave-patterns in IVF. van Gestel, I., IJland, MM., Evers, JL., et al.[2016]

Artificial intelligence (AI) is transforming embryo selection in assisted reproductive technology (ART) by automating the process, which enhances reliability and reproducibility, potentially leading to higher pregnancy rates for couples undergoing IVF.

AI applications in IVF labs are already being developed to assess embryologist performance, ensure quality control, and predict outcomes like embryo viability and live birth rates, suggesting that AI will soon become a standard practice in fertility treatments.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

New frontiers in embryo selection. Glatstein, I., Chavez-Badiola, A., Curchoe, CL.[2023]

An artificial intelligence algorithm was developed to optimize the workflow during IVF by predicting the best day for monitoring ovarian stimulation, achieving a mean error of just 1.355 days in its predictions.

The algorithm also accurately estimated the total number of oocytes retrieved, with an accuracy of 76% using baseline data and 80% when including data from the monitoring day, which can help streamline IVF processes and reduce patient visits.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

An artificial intelligence platform to optimize workflow during ovarian stimulation and IVF: process improvement and outcome-based predictions. Letterie, G., MacDonald, A., Shi, Z.[2022]

A deep learning-based embryo selection model, trained on a dataset of 115,832 embryos from 18 IVF centers, demonstrated strong performance in predicting successful implantation, achieving an area under the curve (AUC) of 0.67 for known implantation data (KID) embryos and 0.95 for all embryos.

The model showed good generalizability across different clinics and patient subgroups, with AUC values ranging from 0.60 to 0.75 for KID embryos, indicating its potential to improve embryo selection in diverse clinical settings while reducing biases associated with manual evaluations.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Robust and generalizable embryo selection based on artificial intelligence and time-lapse image sequences. Berntsen, J., Rimestad, J., Lassen, JT., et al.[2022]

Artificial intelligence classifiers analyzing images of blastocysts can predict the likelihood of achieving a live birth more accurately than conventional embryo evaluation, with overall accuracy rates of 0.72 compared to 0.74 for all age groups combined.

The AI showed improved performance across different maternal age categories, particularly excelling in predicting outcomes for women aged 40-41 and ≥42 years, indicating its potential as a valuable tool in reproductive medicine.

NCBI (Pubmed)

pubmed.ncbi.nlm.nih.gov

Feasibility of deep learning for predicting live birth from a blastocyst image in patients classified by age. Miyagi, Y., Habara, T., Hirata, R., et al.[2023]

References

1. India pubmed.ncbi.nlm.nih.gov

In Contemporary Reproductive Medicine Human Beings are Not Yet Dispensable. [2023]

2. Netherlands pubmed.ncbi.nlm.nih.gov

An ultrasound-based deep learning radiomic model combined with clinical data to predict clinical pregnancy after frozen embryo transfer: a pilot cohort study. [2023]

3. United States pubmed.ncbi.nlm.nih.gov

Complex endometrial wave-patterns in IVF. [2016]

4. Netherlands pubmed.ncbi.nlm.nih.gov

New frontiers in embryo selection. [2023]

5. Netherlands pubmed.ncbi.nlm.nih.gov

An artificial intelligence platform to optimize workflow during ovarian stimulation and IVF: process improvement and outcome-based predictions. [2022]

6. United States pubmed.ncbi.nlm.nih.gov

Robust and generalizable embryo selection based on artificial intelligence and time-lapse image sequences. [2022]

7. Japan pubmed.ncbi.nlm.nih.gov

Feasibility of deep learning for predicting live birth from a blastocyst image in patients classified by age. [2023]