Extended Data Fig. 2: Comparison of VSI-SR models constituted with 2D and 3D convolutional architecture. | Nature Methods

Extended Data Fig. 2: Comparison of VSI-SR models constituted with 2D and 3D convolutional architecture.

From: Fast-adaptive super-resolution lattice light-sheet microscopy for rapid, long-term, near-isotropic subcellular imaging

Extended Data Fig. 2

a–c, Representative images (MIP) acquired and generated by LLSM, conventional LLS-SIM, and VSI-SR models constructed with 2D and 3D convolutional architectures, which were trained by datasets of peroxisomes (a), lysosomes (b), and F-actin (c), respectively. Scale bar, 5 μm (full FOV images), 1 μm (horizontal bars in magnified regions), and 3 μm (vertical bars in magnified regions). d,e, Box-and-whisker plots of the PSNR (d) and SSIM (e) for images reconstructed by VSI-SR models (n = 20 for each specimen). Central line, medians; limits, 75% and 25%; whiskers, maximum and minimum.

Source data

Back to article page