AUTHOR=Lopes Dercilio Junior Verly , Monti Gustavo Fardin , Burgreen Greg W. , Moulin Jordão Cabral , dos Santos Bobadilha Gabrielly , Entsminger Edward D. , Oliveira Ramon Ferreira TITLE=Creating High-Resolution Microscopic Cross-Section Images of Hardwood Species Using Generative Adversarial Networks JOURNAL=Frontiers in Plant Science VOLUME=Volume 12 - 2021 YEAR=2021 URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2021.760139 DOI=10.3389/fpls.2021.760139 ISSN=1664-462X ABSTRACT=Microscopic wood identification plays a critical role in many economically relevant areas in wood science. Historically, producing and curating relevant and representative microscopic cross-section images of wood species is limited to highly experienced and trained anatomists. This paper demonstrates the feasibility of generating synthetic microscopic cross-sections of hardwood species. We leveraged a publicly available dataset of 119 hardwood species to train the StyleGAN generative adversarial network (GAN) to unsupervised learn the probability of normal data. The proposed GAN generated anatomically accurate cross-section images with remarkable fidelity to actual data. Quantitative metrics corroborated the capacity of StyleGAN in capturing complex wood structure by resulting in a Fréchet inception distance (FID) score of 17.38. Image diversity was calculated using the Structural Similarity Index Measure (SSIM). The SSIM results showed that the generated images greatly overlapped ground-truth data distribution, which confirmed that the GAN approach can successfully synthesize diverse images. To confirm the usefulness and realism of the GAN generated images, eight anatomists in two experience levels participated in a visual Turing test (VTT) and correctly identified fake and actual images at rates of 48.3% and 43.7%, respectively, with no statistical difference when compared to random guess (p = 0.671 and 0.064, respectively) when tested at a statistical significance level of p < 0.05. The StyleGAN can synthesize realistic, diverse, and meaningful high-resolution microscope cross-section images that are virtually indistinguishable from real images.