Reconstruction from Multiple Particles for 3D Isotropic Resolution in Fluorescence Microscopy
D. Fortun, P. Guichard, V. Hamel, C.O.S. Sorzano, N. Banterle, P. Gönczy, M. Unser
IEEE Transactions on Medical Imaging, in press.
Please do not bookmark the In Press papers as content and presentation may differ from the published version.
In a few seconds, you should be redirected to the published version. The preprint version is still available here
The imaging of proteins within macromolecular complexes has been limited by the low axial resolution of optical microscopes. To overcome this problem, we propose a novel computational reconstruction method that yields isotropic resolution in fluorescence imaging. The guiding principle is to reconstruct a single volume from observations of multiple rotated particles. Our new operational framework detects particles, estimates their orientation, and reconstructs the final volume. The main challenge comes from the absence of initial template and a priori knowledge about the orientations. We formulate the estimation as a blind inverse problem, and propose a block-coordinate stochastic approach to solve the associated nonconvex optimization problem. The reconstruction is performed jointly in multiple channels. We demonstrate that our method is able to reconstruct volumes with 3D isotropic resolution on simulated data. We also perform isotropic reconstructions from real experimental data of doubly labelled purified human centrioles. Our approach revealed the precise localization of the centriolar protein Cep63 around the centriole microtubule barrel. Overall, our method offers new perspectives for applications in biology that require the isotropic mapping of proteins within macromolecular assemblies.