Fig. 4: 3D SIM experimental results for a RPE-1 cell expressing StayGold-tagged endoplasmic reticulum.
From: Neural space–time model for dynamic multi-shot imaging
a, Maximum projection of the reconstructed volume with color-coded depth. b, Zoom-ins at three time points, each of which corresponds to a different illumination orientation, for the widefield, conventional and NSTM reconstructions. A moving window approach (Methods) is used to compute the conventional reconstruction2 at different time points. The NSTM reconstructions are overlaid with the recovered motion kernels which show the sample’s motion displacements from the previous time point. The colors of the motion kernel indicate motion directions, according to the colorwheel in c. c, Zoom-in NSTM reconstructions at three time points and the combined view with color-coded time. The motion kernels on the second and third time points show the structure’s motion displacements from the previous time point, with color-coding to indicate motion directions.
