Enhancing in vitro maturation with microfluidics and artificial intelligence

In vitro maturation develops immature eggs, or oocytes, in the lab instead of the body. This procedure involves static culture conditions, manual handling, and subjective assessment — limitations that can degrade oocyte quality.

Nguyen et al. present a novel system that combines microfluidics and artificial intelligence-driven image analysis to improve in vitro maturation. Simulation and experiment showed that a specifically designed microfluidic chip with hexagonal micropillars efficiently trapped individual oocytes of various sizes while filtering out smaller oocytes, which are often lower quality.

The chip’s dynamic flow more closely mimics conditions in the body than static cultures, improving nutrient delivery and waste removal around the oocytes. The team demonstrated that it had a higher maturation rate compared to traditional culture methods, suggesting that the chip may provide a more favorable environment for oocyte development.

In addition, image analysis based on a deep learning model allowed the system to automatically evaluate oocyte development at the single-cell level in real time.

“The study demonstrates that combining AI with microfluidic technologies may transform biomedical research by enabling more precise, automated, and physiologically relevant cell culture and analysis systems,” said author Thu Hang Nguyen. “I think this platform could play an important role in optimizing in vitro maturation and advancing the next generation of assisted reproductive technologies.”

Next, the authors plan to optimize the system and validate its performance on animal models before it is translated to human applications.

“Our research team collaborated closely with clinicians and biomedical research institutes, which enabled us to work towards practical and human-centered applications in assisted reproductive health,” Nguyen said.

Source: “FEMI: deep learning - Assisted analysis of single oocytes trapping and maturation for enhanced on-chip IVM,” Thu Hang Nguyen, Tung Thanh Le, Hanh Van Nguyen, Hang Thu Bui, Hoang Anh Phan, Tung Thanh Bui, Trinh Chu Duc, and Loc Quang Do, Biomicrofluidics (2026). The article can be accessed at https://doi.org/10.1063/5.0328820 .