Blind deblurring in fluorescence microscopy
Jointly estimating the sharp image and the unknown blur from a single fluorescence microscopy image, with deep learning and diffusion priors.
- Covers diffraction-limited blurs (Zernike polynomials), spatially varying blurs, and 2D/3D settings.
- Efficient PyTorch implementation, with ongoing validation on RIM, TIRF and confocal microscopes.
- Theory: a Morse-Bott analysis of MAP estimation with learned priors, explaining when gradient-based methods recover the right image and why the “blurry trap” persists.
- Joint work with Pierre Weiss, Florian Sarron, Paul Escande, Thomas Mangeat and Sylvain Cantaloube. Presented at MIFOBIO 2025.