BACK ON FRIDAYS after the Virtual Pub!
We are moving our Tech Exchange back to Fridays 2pm CE(S)T after the Euro-BioImaging Virtual Pub to bring together even more experts from the imaging industry and academia in an exciting two hours of imaging technology talks!
Same Zoom link, same crowd!
In episode #9 we hear about image data viewing, analysis and AI using open-source software and how commercial software can learn from users!
Tom Herolds from scalableminds will speak about “Analyzing and reconstructing volumetric image datasets with webKnossos“.
Learning from users how to gain insights from microscopy images – a case study for applied AI towards autonomous image analysis.
Luciano Lucas, PhD – Director Data & Analysis, Leica Microsystems
“I do the same type of image analysis all the time, why doesn’t the software just learn how I do it?” That is the question the Data and Analysis team at Leica Microsystems started with. We have now built an integrated system which passively learns from user behavior while they are working. The system gathers telemetry data as the user interacts with Aivia. The telemetry consists of key actions, the used image analysis parameters and a compressed, processed, anonymized version of the image data. With this data we have trained an Artificial Intelligence – using state of the art neural networks – capable of predicting which image analysis parameters (e.g. background removal, threshold level, typical object size, partition settings, filtering) are needed to correctly process images the system has never seen. We will discuss what data and neural networks we have used to achieve autonomous image analysis. We will also explore the pitfalls and limitations of this system as well as future developments.