Wednesday, 12th of February 2025 at 9:00 am
Chairs: Sabine Baumgart & Claudia Peitzsch
Benjamin Ehret
Data Scientist, Navignostics AG
Enhancing IMC through AI-based information transfer between antibody panels
The emergence of multiplexed protein imaging technologies, such as imaging mass cytometry, has enabled the analysis of tumor tissues at an unprescedented level of detail. However, the number of proteins of interest still exceeds the number of available imaging channels. One way to increase the number of analyzed proteins is to image consecutive tissue slices with different antibody panels. Yet, this approach has the limitation that one cannot analyze the spatial interactions of cells stained with the different panels. To address this issue, we use deep learning to transfer information between the two multiplexed images. In this talk, I will present an application of this approach to a colorectal cancer data set, imaged with a tumor panel and an immune panel. We use the two panels to define different sets of cell types and show that we can train a neural network to infer immune cell types in the tumor panel image and vice versa. This information transfer allows us to perform spatial analysis of detailed immune and tumor cell types, leading to a more comprehensive understanding of the analyzed tissues.
Biosketch
Dr. Benjamin Ehret is a Data Scientist at Navignostics AG, a University of Zurich spin-off specializing in spatial single-cell proteomics for personalized cancer treatment. He joined the company in September 2023, bringing expertise in image and data analysis, as well as machine learning, to support the company’s mission in advancing cancer care and drug development.
Benno earned his PhD in Neuroscience from ETH Zürich in 2022, where he focused on the role of the prefrontal cortex in learning stimulus-response mappings and developed novel methods for interpreting high-dimensional neural activity recordings. He also researched continual learning in recurrent neural networks. Prior to this, he completed his Master of Science in Neural Systems and Computation at ETH Zürich and the University of Zurich and his Bachelor of Science in Bioinformatics at the Ludwig-Maximilians-Universität München and the Technical University of Munich.