Wave-Reconstruction Generative Adversarial Networks (WRGAN): wave function predicted from a high-resolution electron microscopy images

Abstract

High-resolution transmission electron microscopy (HRTEM) is indispensable for atomic-scale characterization yet fundamentally limited by the inherent phase loss in conventional detectors including CCD. To overcome this barrier, we propose Wave-Reconstruction Generative Adversarial Networks (WRGAN) that directly predict wave function amplitude and phase from single HRTEM images. Our physics-guided framework employs a Unet++ generator within a Generative Adversarial Networks (GAN) architecture via defining a physics-guided consistency loss. A key advantage is that WRGAN, trained solely on simulated data, demonstrates robust performance when directly applied to experimental images. Validation on experimental Nb₈W₉O₄₇ image shows predicted amplitudes and phases closely match the groundtruth wave functions. Significantly, WRGAN successfully resolves upper and lower surface projections in noisy single-wall carbon nanotube (SWCNT) images, enabling near-atomic-resolution 3D reconstruction.

Keywords

High-resolution transmission electron microscopy / deep learning / carbon nanotube / phase retrieval / generative adversarial networks

References

[1] Xu T.,Hou L.,Zhou Z..et al. Lithiation dynamics in vertically aligned 1T-PtSe₂ and its application in dendrite-free lithium metal batteries Appl. Phys. Lett. 2025 126 203901

[2] Zhang B.,Zhao L.,Cheng Y.,Golberg D.,Wang M. S.. Reversible tuning of individual carbon nanotube mechanical properties via defect engineering Nano Lett. 2016 16 5221 7

[3] Vincent J. L.,Crozier P. A.. Atomic level fluxional behavior and activity of CeO₂-supported Pt catalysts for CO oxidation Nat. Commun. 2021 12 5789 PMC8490411

[4] Li X.,Wang J.,Liu X..et al. Direct imaging of tunable crystal surface structures of MOF MIL-101 using high-resolution electron microscopy J. Am. Chem. Soc. 2019 141 12021 8

[5] Han X.,Liu P.,Lin F..et al. Structures and structural evolution of sublayer surfaces of metal-organic frameworks Angew. Chem. Int. Ed. Engl. 2020 59 21419 24

[6] Guedj C.,Hung L.,Zobelli A.,Blaise P.,Sottile F.,Olevano V.. Evidence for anisotropic dielectric properties of monoclinic hafnia using valence electron energy-loss spectroscopy in high-resolution transmission electron microscopy and ab initio time-dependent density-functional theory Appl. Phys. Lett. 2014 105 222904

[7] Hsieh W. K.,Chen F. R.,Kai J. J.,Kirkland A. I.. Resolution extension and exit wave reconstruction in complex HREM Ultramicroscopy 2004 98 99 114

[8] Bhat U.,Datta R.. Direct methods applied to phase retrieval in high resolution transmission electron microscopy J. Phys. Commun. 2022 6 045007

[9] Coene W.,Thust A.,Op De Beeck M.,Van Dyck D.. Maximum-likelihood method for focus-variation image reconstruction in high resolution transmission electron microscopy Ultramicroscopy 1996 64 109 35

[10] Allen L. J.,McBride W.,O’Leary N. L.,Oxley M. P.. Exit wave reconstruction at atomic resolution Ultramicroscopy 2004 100 91 104

[11] Ming W. Q.,Chen J. H.,He Y. T.,Shen R. H.,Chen Z. K.. An improved iterative wave function reconstruction algorithm in high-resolution transmission electron microscopy Ultramicroscopy 2018 195 111 20

[12] Tang C. Y.,Chen J. H.,Zandbergen H. W.,Li F. H.. Image deconvolution in spherical aberration-corrected high-resolution transmission electron microscopy Ultramicroscopy 2006 106 539 46

[13] Lehtinen O.,Geiger D.,Lee Z..et al. Numerical correction of anti-symmetric aberrations in single HRTEM images of weakly scattering 2D-objects Ultramicroscopy 2015 151 130 5

[14] Lin F.,Ren X. B.,Zhou W. P..et al. Exit-wave phase retrieval from a single high-resolution transmission electron microscopy image of a weak-phase object Micron 2018 114 23 31

[15] Zhang X.,Chen S.,Wang S.,Huang Y.,Jin C.,Lin F.. Exit wave reconstruction of a focal series of images with structural changes in high-resolution transmission electron microscopy J. Microsc. 2024 296 24 33

[16] Meng Y.,Wang S.,Ren X..et al. Combining machine learning algorithms with traditional methods for resolving the atomic-scale dynamic structure of monolayer MoS₂ in high-resolution transmission electron microscopy Chinese Phys. B. 2025 34 016802

[17] Ede J. M.,Peters J. J. P.,Sloan J.,Beanland R.. Exit wavefunction reconstruction from single transmission electron micrographs with deep learning. arXiv 2020;arXiv:2001.10938. Available online: https://arxiv.org/abs/2001.10938. (accessed 2026-4-3)

[18] Lee J.,Lee Y.,Kim J.,Lee Z.. Contrast transfer function-based exit-wave reconstruction and denoising of atomic-resolution transmission electron microscopy images of graphene and Cu single atom substitutions by deep learning framework Nanomaterials (Basel). 2020 10 1977 PMC7601262

[19] Leth Larsen M. H.,Dahl F.,Hansen L. P..et al. Reconstructing the exit wave of 2D materials in high-resolution transmission electron microscopy using machine learning Ultramicroscopy 2023 243 113641

[20] Meng Z.,Ming W.,He Y.,Shen R.,Chen J.. Exit wave function reconstruction from two defocus images using neural network Micron 2024 177 103564

[21] Goodfellow I.,Pouget-abadie J.,Mirza M..et al. Generative adversarial networks Commun. ACM 2020 63 139 44

[22] Zhou Z.,Siddiquee M. M. R.,Tajbakhsh N.,Liang J.. UNet++: A Nested U-Net Architecture for Medical Image Segmentation. arXiv 2018;arXiv:1807.10165. Available online: https://arxiv.org/abs/1807.10165. (accessed 2026-4-3)

[23] Srivastava N.,Hinton G.,Krizhevsky A.,Sutskever I.,Salakhutdinov R.. Dropout: a simple way to prevent neural networks from overfitting J. Mach. Learn. Res. 2014 15 1929 58

[24] Ioffe S.,Szegedy C.. Batch normalization: accelerating deep network training by reducing internal covariate shift. arXiv 2015;arXiv:1502.03167. Available online: https://arxiv.org/abs/1502.03167. (accessed 2026-4-3)

[25] Maas A. L.,Hannun A. Y.,Ng A. Y.. Rectifier nonlinearities improve neural network acoustic models Proc. ICML 2013 3 , 1-6. https://www.awnihannun.com/papers/relu_hybrid_icml2013_final.pdf (accessed 2026-4-3)

[26] Dumoulin V.,Visin F.. A guide to convolution arithmetic for deep learning. arXiv 2016;arXiv:1603.07285. Available online: https://arxiv.org/abs/1603.07285. (accessed 2026-4-3)

[27] Zhao H.,Liu F.,Li L.,Luo C.. A novel softplus linear unit for deep convolutional neural networks Appl. Intell. 2017 48 1707 20

[28] Meyer R. R.,Kirkland A. I.,Saxton W. O.. A new method for the determination of the wave aberration function for high resolution TEM 1. Measurement of the symmetric aberrations Ultramicroscopy 2002 92 89 109

[29] Yuan P. J.,Wu K. P.,Chen S. W..et al. ToTEM: a software for fast TEM image simulation J. Microsc. 2022 287 93 104