Inverse design of unparametrized nanostructures by generating images from spectra.

Recently, there has been an increasing number of studies applying machine learning techniques for the design of nanostructures. Most of these studies train a deep neural network (DNN) to approximate the highly nonlinear function of the underlying physical mapping between spectra and nanostructures. At the end of training, the DNN allows an on-demand design of nanostructures, i.e., the model can infer nanostructure geometries for desired spectra. While these approaches have presented a new paradigm, they are limited in the complexity of the structures proposed, often bound to parametric geometries. Here we introduce spectra2pix, which is a DNN trained to generate 2D images of the target nanostructures. By predicting an image, our model architecture is not limited to a closed set of nanostructure shapes, and can be trained for the design of a much wider space of geometries. We show, for the first time, to the best of our knowledge, a successful generalization ability, by designing completely unseen shapes of geometries. We attribute the successful generalization to the ability of a pixel-wise architecture to learn local properties of the meta-material, therefore mimicking faithfully the underlying physical process. Importantly, beyond synthetical data, we show our model generalization capability on real experimental data.

Variable-Density Single-Shot Fast Spin-Echo MRI with Deep Learning Reconstruction by Using Variational Networks.

Purpose To develop a deep learning reconstruction approach to improve the reconstruction speed and quality of highly undersampled variable-density single-shot fast spin-echo imaging by using a variational network (VN), and to clinically evaluate the feasibility of this approach. Materials and Methods Imaging was performed with a 3.0-T imager with a coronal variable-density single-shot fast spin-echo sequence at 3.25 times acceleration in 157 patients referred for abdominal imaging (mean age, 11 years; range, 1-34 years; 72 males [mean age, 10 years; range, 1-26 years] and 85 females [mean age, 12 years; range, 1-34 years]) between March 2016 and April 2017. A VN was trained based on the parallel imaging and compressed sensing (PICS) reconstruction of 130 patients. The remaining 27 patients were used for evaluation. Image quality was evaluated in an independent blinded fashion by three radiologists in terms of overall image quality, perceived signal-to-noise ratio, image contrast, sharpness, and residual artifacts with scores ranging from 1 (nondiagnostic) to 5 (excellent). Wilcoxon tests were performed to test the hypothesis that there was no significant difference between VN and PICS. Results VN achieved improved perceived signal-to-noise ratio (P = .01) and improved sharpness (P < .001), with no difference in image contrast (P = .24) and residual artifacts (P = .07). In terms of overall image quality, VN performed better than did PICS (P = .02). Average reconstruction time ± standard deviation was 5.60 seconds ± 1.30 per section for PICS and 0.19 second ± 0.04 per section for VN. Conclusion Compared with the conventional parallel imaging and compressed sensing reconstruction (PICS), the variational network (VN) approach accelerates the reconstruction of variable-density single-shot fast spin-echo sequences and achieves improved overall image quality with higher perceived signal-to-noise ratio and sharpness. © RSNA, 2018 Online supplemental material is available for this article.

Plasmonic nanostructure design and characterization via Deep Learning.

Nanophotonics, the field that merges photonics and nanotechnology, has in recent years revolutionized the field of optics by enabling the manipulation of light-matter interactions with subwavelength structures. However, despite the many advances in this field, the design, fabrication and characterization has remained widely an iterative process in which the designer guesses a structure and solves the Maxwell's equations for it. In contrast, the inverse problem, i.e., obtaining a geometry for a desired electromagnetic response, remains a challenging and time-consuming task within the boundaries of very specific assumptions. Here, we experimentally demonstrate that a novel Deep Neural Network trained with thousands of synthetic experiments is not only able to retrieve subwavelength dimensions from solely far-field measurements but is also capable of directly addressing the inverse problem. Our approach allows the rapid design and characterization of metasurface-based optical elements as well as optimal nanostructures for targeted chemicals and biomolecules, which are critical for sensing, imaging and integrated spectroscopy applications.