Fig. 5: Zero-shot denoising and resolution enhancement in multimodal SIM data.
From: Zero-shot learning enables instant denoising and super-resolution in optical fluorescence microscopy
a Schematic of the training procedure of ZS-DeconvNet for SIM. b Progression of SNR and resolution improvement across the CCPs in a SUM-159 cell, from raw SIM images (left), conventional SIM image (right), and ZS-DeconvNet enhanced SIM image (middle). c Progression of SNR and resolution improvement across the microtubules in a COS-7 cell, from raw SIM images (left), conventional SIM image (right), and ZS-DeconvNet enhanced SIM image (middle). d Representative maximum intensity projection (MIP) images of F-actin in a HeLa cell obtained via LLSM, LLS-SIM, and LLS-SIM enhanced by 3D ZS-DeconvNet across three dimensions. e, Representative MIP images of mitochondrial outer membrane labeled with TOMM20 in a 293 T cell obtained via LLSM, LLS-SIM, and LLS-SIM enhanced by 3D ZS-DeconvNet across three dimensions. Scale bar, 1 μm (a), 2 μm (b, c), 0.5 μm (zoom-in regions in b, c), 3 μm (d, e).
