A New Era - Self-learning AI for Microscopy
Discover the latest developments in artificial intelligence (AI), deep learning and convolutional neural networks for live cell analysis.
There has been an unprecedented run of recent breakthroughs in artificial intelligence for vision applications, driven by deep neural networks (DNNs) and deep learning. However, the practical application of AI in microscopy has been challenging due to insufficient robustness and inherently long training phases associated with neural networks.
To overcome this, Olympus has developed self-learning microscopy on the scanR high content screening system. The microscope automatically generates the ‘ground truth’ data required for training DNNs by acquiring reference images during the training phase.
Training DNNs to perform image analysis tasks robustly—even under challenging conditions— is therefore easy with the Olympus scanR system and requires minimal human supervision. The performance of these DNNs often exceeds traditional approaches and opens doors to new life science microscopy applications.
Self-learning microscopy has many benefits, including automated analysis of cell populations in microwell plates without the need for fluorescent labeling. AI can reliably detect and analyze cell nuclei from brightfield images with an accuracy that exceeds fluorescence-based methods.
An upcoming Olympus webinar will detail examples of how easy it is to train DNNs using the scanR AI-based high content screening system as well as explore the latest developments and benefits of AI for microscopy. Join the webinar to learn more (2 April, 10–11am CEST): www.olympus.eu/deep-learning-webinar