New Perspectives in Optical Microscopy - How Artificial Intelligence Speeds up your Experiment Startup

Martin Gleisner, Ph.D.

Carl Zeiss Microscopy GmbH

Dr. Martin Gleisner grew up in Weimar, Germany, and studied chemistry in Göttingen. From the beginning, he was fascinated with fluorescence microscopes and their numerous applications and decided to pursue his Ph.D. in the interdisciplinary field of biophysical chemistry. In doing so, he spent considerable time optimizing microscopic setups to make them more suitable for biological applications. Before he joined ZEISS, he held a position at Analytik Jena AG in the area of elemental analysis. In his role in the Life Sciences business sector, he is mainly responsible for widefield microscopes and accessories.


Optical microscopy has been a valuable tool in life science research for many decades. Beginning with a widefield setup, many more microscopy techniques have been developed over time. These techniques made it possible to analyze samples faster, in more dimensions, and with higher resolution. All this led to new fields of application and many scientific breakthroughs. What all these developments have in common is that the improved performance could only take effect after the sample was found and brought into focus. Even with all the automation and motorization in state-of-the-art optical microscopes in life science research, the method for finding the sample did not change significantly over time. It still requires a trained operator with microscope knowledge and is mostly done manually by looking through the eyepieces and using the focus and stage control for adjustments.
The latest developments in artificial intelligence have enabled the design of a tool that eliminates these manual steps. The AI Sample Finder automatically detects the sample carrier, adjusts the focus, and can detect even low-contrast, unstained samples. This automation results in an overview image for fast and convenient navigation that will benefit both beginners and experts. The time to image is reduced by up to a factor of 10, improving throughput and ease of use.