Program of ISCS Workshop

Lecture 1: Mathematics of Computational Imaging Systems, by Nelly Pustelnik (ENS Lyon, France)

1- Introduction to inverse problems in imaging

2- From sparsity-driven proximal algorithms to deep learning
2.1- Sparsity in image processing
2.2- Proximal algorithms for convex optimization
2.3- Proximal algorithms for nonconvex optimization
2.4- Neural networks and its link to proximal algorithms

3- Model based neural networks
3.1- Plug-and-play algorithms
3.2- Unfolded neural network

Lecture 2: Physics of Computational Imaging Systems, by Gail McConnell (Uni. of Strathclyde, Scotland)

1- Optical image formation

2- Light sources and detectors in optical microscopy

3- Contrast in optical microscopy

4- Advantages and disadvantages of labelling for optical microscopy

5- Common microscopy methods in optical microscopy

6- Current trends and opportunities in optical microscopy

7- Multiscale imaging

8- Considerations for designing an optical microscope, with emphasis on the Mesolens

9- Optical mesoscopy and its applications in biology and biomedicine

Track 1: DeepInverse: A Pytorch Library For Solving Imaging Inverse Problems With Deep Learning, by Julian Tachella (ENS Lyon, France), Matthieu Terris (INRIA, France), and Samuel Hurault (CNRS, France)

1- Introduction to the library and forward operators

2- Presentation of mini-projects

3- Iterative methods: explicit regularizations (TV, wavelets, etc.), and implicit regularizations (plug-and-play)

4- Sampling methods: Plug-and-play ULA and diffusion methods

5- Training: supervised and self-supervised losses, training details

6- Division of participants in small groups per mini-project

7- Coding session with organizers providing support

8- Short presentation of mini-projects by participants

Track 2: Build Your Own Radar with MIT’s Next-Generation Kit, by Bradley T. Perry and Kenneth E. Kolodziej (MIT, USA)