Fig. 2: EMDiffuse achieves superior performance in super-resolution. | Nature Communications

Fig. 2: EMDiffuse achieves superior performance in super-resolution.

From: Diffusion-based deep learning method for augmenting ultrastructural imaging and volume electron microscopy

Fig. 2

a Schematic of EMDiffuse-r for super-resolution of EM images. b Representative region from mouse brain sections (input with a pixel size of 6.6 nm, GT with a pixel size of 3.3 nm) for comparing the super-resolution capability of EMDiffuse-r with CARE, PSSR, and RCAN. Top right of each panel is the Fourier power spectrum. The resolution of processed images is shown on the bottom left of the image. The uncertainty value is shown in the top left corner. c The 3D scatter plot and distribution plots show the quantitative performance assessment of EMDiffuse, CARE, PSSR, and RCAN for the super-resolution task. Each point represents a unique test image of the mouse brain cortex, n = 960. d The Fourier ring correlation plot between predictions in (b) and ground truth image. e EMDiffuse for super-resolution is robust and transferable to other tissues, including the liver, heart, and bone marrow. Resolution values and uncertainty values are indicated on each panel. U uncertainty value. Scale bar, b 0.1 μm; d 0.3 μm. Source data are provided as a Source Data file.

Back to article page