Imaging lies at the heart of the investigation of biological processes from the molecular level to entire organisms. While steady advances in microscopy technology are opening new windows into the inner workings of the biological systems, imaging has hard limits stemming from optics of the microscope, chemistry of fluorophores and the necessity to keep the imaged sample alive. Recently, deep learning has emerged as a powerful approach to making trade-offs between imaging speed, image quality and sample health, that make many previously inaccessible imaging scenarios possible. It has been shown that by incorporating deep learning driven image restoration into imaging pipelines, it is possible to image biological specimen with dramatically reduced laser power, achieve isotropic resolution from under-sampled 3D volumes and even resolve structures below the diffraction limit faster.

This course will focus on the application of leading deep learning framework for microscopy image restoration, the Content Aware Image Restoration (CARE), to various biological image restoration tasks. The power of deep learning in microscopy will be demonstrated using open source tools such as Fiji, KNIME and Jupyter notebooks that make the advanced technologies accessible to anyone.

Besides the in-depth practical experience with CARE, the participants will learn about alternative computational approaches to super-resolution microscopy and about handling of big microscopy image datasets. Finally, the impact of computational techniques on biological imaging will be demonstrated on frontier biological applications spanning across scales of biological complexity from molecular (Cryo-EM), to subcellular (nuclear organization) to organismal (tissue mechanics) levels.

During the practical part of the course the students will learn how to:

• Collect appropriate training data on the provided samples using diverse microscopy hardware.
  
• Train CARE networks on state-of-the-art as well as commodity computer hardware.
• Apply CARE networks to restore the data acquired during the course.
• Learn how to evaluate the CARE results and avoid potential artefacts.