ENDOSCAN : AI for Endometrium at ASRM
We’ve just returned from the ASRM 2025 Congress in Texas, USA, where our team had the pleasure of presenting two oral presentations on advancing endometrial assessment with artificial intelligence:
- Endometrial Thickness Measurement in IVF: Comparing Clinician and AI Performance
- Automated Endometrial Volume Measurement from 3D Ultrasound Using Deep Learning
Both studies showcased the power of AI to bring greater precision, consistency and efficiency to one of the most challenging aspects of fertility care – endometrial evaluation.
O-131 – ENDOMETRIAL THICKNESS MEASUREMENT IN IN VITRO FERTILIZATION (IVF): COMPARING CLINICIAN AND ARTIFICIAL INTELLIGENCE (AI) PERFORMANCE IN ULTRASOUND ASSESSMENT
Our first oral presentation explored how artificial intelligence can assist clinicians in measuring endometrial thickness during IVF.
In a multi-center study spanning both Poland and the United States, we analyzed over 150 ultrasound videos captured during IVF cycles. The findings were striking: our AI system achieved a level of measurement consistency comparable to expert clinicians, matching their accuracy within just over 1 mm on average.
Even more promising, when evaluating a clinically relevant threshold of 7 mm (often used as a criterion for endometrial receptivity) the AI’s agreement with clinicians exceeded 89%, slightly higher than the agreement between clinicians themselves.
In other words, AI not only measured as accurately as humans but also brought an added layer of standardization – a quality often missing in ultrasound-based assessments that depend heavily on individual operator experience.
These results demonstrate how tools like ENDOSCAN can reduce observer-dependent variability, offering fertility specialists more reliable data for decision-making and smoother collaboration across centers.
O-132 – AUTOMATED ENDOMETRIAL VOLUME MEASUREMENT FROM 3D ULTRASOUND USING DEEP LEARNING – AI in Endometrium
Our second study took AI’s potential a step further, focusing on automating endometrial volume measurement from 3D ultrasound scans.
Using a deep learning model trained on over 4,000 annotated images from clinics across three continents, we validated an algorithm capable of producing precise volume estimates in under two seconds – compared to up to ten minutes of manual work required by traditional methods.
The model’s performance closely mirrored that of the clinical gold standard (VOCAL), with an impressive correlation of 0.95 between AI and expert measurements. More than 90% of results were within just 1 cm³ of the manually measured values.
This leap in efficiency and precision opens the door to standardized, large-scale analyses of endometrial dynamics throughout the stimulation cycle – something that has long been limited by the time-intensive nature of manual 3D assessment.
ENDOSCAN: From Research to Real-World Practice
These research efforts are part of the foundation on which we’ve built ENDOSCAN, our AI-powered tool designed to transform how clinicians assess the endometrium. It aims to eliminate subjectivity from ultrasound interpretation, streamline workflows, and enhance decision-making in IVF cycles.
ENDOSCAN supports medical professionals with automated, standardized, and reproducible measurements, helping ensure that every patient receives the best-informed care possible.
And this is just the beginning. Together with FOLLISCAN, our complementary solution for ovarian monitoring, ENDOSCAN forms part of our growing ecosystem of AI tools dedicated to reproductive medicine innovation.
Stay tuned for more updates as we continue testing, refining, and expanding ENDOSCAN to meet the evolving needs of fertility specialists worldwide.
You can discover more information about ENDOSCAN here: https://mimfertility.ai/endoscan/
