A Hybrid Closed-loop Droplet Detection for Digital Microfluidics

Aiming at the key technical difficulties of droplet detection in digital microfluidic chips, we propose a closed-loop feedback system combining electrical impedance and image sensing technology. The system integrates the image processing algorithm based on YOLOv5 and impedance sensing technology to achieve accurate detection of droplets. With only 500 training samples, the detection accuracy of the YOLOv5 reaches over 95%, which improves the efficiency and accuracy of droplet recognition. In addition, we develop a self-adjusting mechanism that can automatically plan the path and guide the target droplet to effectively bypass the faulty electrode and accurately reach the target position. The success rate reached 100% in 50 tests, which greatly enhanced the stability and reliability of the system. Through electrical impedance detection, the quantitative relationship between the droplet volume and the detection value was further explored. The single-electrode droplet volume detection time is only about 1.5 ms, and the effective detection resolution is 0.15 μL. This work achieved a deep combination of electrical impedance and image sensing technology in droplet detection, and optimized the digital microfluidic droplet detection system. It not only provided a more reliable and stable experimental platform for biochemical experiments, but also provided new ideas and directions for the development of intelligent microfluidic detection systems and the realization of automated and intelligent droplet detection and analysis.