Lab Automation Support ～Integrating image-analysis AI into automated medium-scale synthesis systems～

Challenge

Medium-scale synthesis experiments are needed in the lead up to commercial-scale drug production. As part of its laboratory automation initiative, Daiichi Sankyo Co., Ltd. was developing an automated medium-scale synthesis system. However, conventional image processing systems using camera sensors faced challenges such as the false detection of reaction solutions caused by residue on vessel surfaces, and difficulty identifying liquid interfaces in opaque solutions. Additionally, there was a growing need to acquire data suitable for scaling-up and for automating experimental operations. In collaboration with Daiichi Sankyo, LPIXEL developed an AI-based image analysis technology capable of accurately detecting liquid levels and interfaces under diverse conditions.

LPIXEL’s Solution – AI Technology and Impact

By applying AI-based image analysis, liquid levels and interfaces inside reactors can be accurately detected, enabling precise volume measurement and reliable interface detection during liquid separation. To ensure robust performance in practical use, the AI model was trained to handle a wide range of conditions. As a result, it maintains high accuracy even in challenging situations involving foaming, residue on vessel surfaces, or variations in solution color—issues that conventional image-processing systems struggled to overcome. The introduction of this technology is expected to improve experimental reproducibility, reduce variability caused by manual operations, and enhance overall process efficiency while reducing the workload of researchers.