Skip to main content
. 2020 Apr 22;11:1934. doi: 10.1038/s41467-020-15784-x

Fig. 2. Super-resolution imaging under extreme low-light conditions.

Fig. 2

a Two U-Nets were stacked through skip-layer connections. Fifteen SIM raw data images taken under low-light conditions were used as the input and the corresponding SIM reconstructions under normal-light conditions were used as the ground truth to train the scU-Net. b Reconstruction results for different subcellular structures (first row: microtubules; second row: adhesions; third row: mitochondria; fourth row: F-actin). Shown are average projections of 15 SIM raw data (first column), the reconstruction results from a conventional SIM reconstruction algorithm (second column), U-Net-SIM15 output (third column), scU-Net output (fourth column), and the ground truth from SIM reconstruction under normal-light conditions (fifth column). Shown are representative images randomly selected form the testing dataset indicated in Supplementary Table 1. The training datasets were collected from at least three independent experiments. The local enlargements show the restoration quality. Scale bar: 1 μm.