Skip to main content
. 2022 Oct 31;19(11):1427–1437. doi: 10.1038/s41592-022-01652-7

Extended Data Fig. 7. RLN provides better generalization on super-resolution data than other networks.

Extended Data Fig. 7

a) Super-resolved images of live U2OS cells expressing mEmerald-Tomm20-C-10, acquired with iSIM. Top: raw input; bottom: RLN output. b) Higher magnification view of yellow rectangular region in a), comparing ground truth, RLN output, and DDN output. The models were trained with ER datasets. c) SSIM and PSNR measurements for RLN, CARE, RCAN and DDN for data shown in a), means and standard deviations are obtained from N = 6 volumes. d) Super-resolved images of live U2OS cells expressing Lamp1-EGFP, acquired with iSIM. Top: raw input; bottom: RLN output. e) Higher magnification of rectangular regions in d), comparing ground truth, RLN output, and DDN output. Models were trained with phantom objects consisting of dots, solid spheres, and ellipsoidal surfaces. f) SSIM and PSNR measurements, comparing RLN, CARE, RCAN and DDN for data shown in d), means and standard deviations are obtained from N = 6 volumes (open circles indicate individual values). Scale bars: a, d) 5 μm, b, e) 2 μm.