At Centre for Cellular Imaging, (CCI) microscopy experts now work side by side with computer scientists to accelerate the implementation of Smart Microscopy, thus enhancing its support to researchers.
Though Smart Microscopy technology has been around for some years now the momentum in the development of the technology has increased in the last five years.
“The first-generation smart microscopy could analyse images and had the ability to stop and ask for new images in case they are out-of-focus. Computers can now not only ask for new ones if the image quality is too low but also suggest the next step in an ongoing experiment. It adapts along the way,” says Simon Leclerc, Associate Researcher, and microscopy expert at CCI.
“Long term, Smart Microscopy could potentially allow researchers and computers to ‘discuss’ how to proceed with a study. Another clear advantage is that the computer support minimizes the risk of confirmation bias,” Simon continues.
At CCI, the expert team holds multiple competencies and team members complete each other when adopting new technology and developing Smart Microscopy.
“Smart Microscopy technology requires giant databases, and it would be great if research funding could address that need. In the future I also hope that industry and universities collaborate closely and overcome the obstacles for sharing knowledge,” says Anders Folkesson, Research Engineer and computer scientist at CCI.
The computers gather and compile information from scientific publications where smart microscopy has been involved. However, language models cannot currently detect the exact context behind an image, for example how the image was acquired or the type of label used, details that are needed to draw the right conclusions. Nor can the language model identify the most recent results, but standardisations are underway. High-capacity databases provide more detailed information though, which in turn means that language models can be better trained. An important enabler for the development of Smart Microscopy is therefore a high availability of research results.
Simon agrees with Anders that availability is crucial.
“A key success factor for Smart Microscopy is that the so-called FAIR principles are applied to a greater extent. FAIR stands for Findable, Accessible, Interoperable and Reusable. In short, that research data are easy to find and that there is information on how to access them. Also, that they are compatible to other data and possible to reuse. Being in the midst of the Smart Microscopy development and experiencing being aware of its potential, I also wish that investments in databases would be addressed. In the long term, smart microscopy is relaying on that, because of the enormous amount of data generated.
