Innovative deep learning approaches enable quantitative analyses of synaptic imaging datasets (C10#)

• Subject Area: Molecular Biology and Physiology of Neurons and Glial Cells
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 317475864

Scientists in the CRC use advanced imaging methods to study synapses and obtain parameters for synaptic modeling, which are currently obtained through time-consuming manual analysis. Deep learning-based approaches could automate these tasks, but requires substantial amounts of annotated training data and struggle with generalization to different datasets. To overcome these limitations, I plan to develop methods that build upon recent advances in domain adaptation and self-supervised learning, and will use them to address three of the most challenging image analysis problems in the CRC. The methods will enable future applications in several microscopy domains, representing a significant advancement for image analysis in biology.

• DFG Programme: Collaborative Research Centres
• Subproject of: SFB 1286: Quantitative Synaptology
• Applicant Institution: Georg-August-Universität Göttingen
• Project Head: Professor Dr. Constantin Pape