Olympus' newly launched scanR 3.1 high-content screening (HCS) station fully embraces the capabilities of artificial intelligence (AI) to enable cutting-edge life science research. It combines the modularity and flexibility of a microscope-based setup with the automation, speed, throughput and reproducibility of HCS applications. Using the 'self-learning microscopy' concept, scanR 3.1 makes it easy to gather data quickly from large live cell populations for reliable, well-supported experimental results.
To minimise setup time, the self-learning approach makes use of a short, one-time training phase in which the software uses a quickly acquired set of images to generate 'ground truth' data without requiring human annotations. It then uses convolutional neural networks to autonomously create robust algorithms that can rapidly analyse large sets of images.
One application that clearly shows the power of AI in HCS is label-free quantification of live cells. Olympus' scanR HCS software can reliably derive nuclei positions in micro wells solely from brightfield transmission images with an accuracy that rivals fluorescence. Quantifying live cells from brightfield images instead of fluorescence shortens exposure times, avoids using genetic modifications or nucleus markers, and saves fluorescence channels for other markers. These benefits reduce phototoxicity and lead to simpler, faster image acquisition and better cell viability.
Reliable low light analysis has also become possible thanks to scanR's AI-based imaging software. Olympus has shown that its software accurately detects DAPI-labelled cells at only 0.2% of the optimal light intensity. It can even distinguish different stages of the cell cycle, based on the intensity of the signal, providing better insights and improving reproducibility.
