gevero

We present a novel approach based on machine learning for designing photonic structures. In particular, we focus on strong light confinement that allows the design of an efficient free-space-to-waveguide coupler which is made of Si- slab overlying on the top of silica substrate. The learning algorithm is implemented using bitwise square Si- cells and the whole optimized device has a footprint of $\boldsymbol{2 \, \mu m \times 1\, \mu m}$, which is the smallest size ever achieved numerically. To find the effect of Si- slab thickness on the sub-wavelength focusing and strong coupling characteristics of optimized photonic structure, we carried out three-dimensional time-domain numerical calculations. Corresponding optimum values of full width at half maximum and coupling efficiency were calculated as $\boldsymbol{0.158 \lambda}$ and $\boldsymbol{-1.87\,dB}$ with slab thickness of $\boldsymbol{280nm}$. Compared to the conventional counterparts, the optimized lens and coupler designs are easy-to-fabricate via optical lithography techniques, quite compact, and can operate at telecommunication wavelengths. The outcomes of the presented study show that machine learning can be beneficial for efficient photonic designs in various potential applications such as polarization-division, beam manipulation and optical interconnects.

Nature Photonics 11, 70 (2017). doi:10.1038/nphoton.2017.2

Authors: Lin Zhou, Yingling Tan, Jingyang Wang, Weichao Xu, Ye Yuan, Wenshan Cai, Shining Zhu & Jia Zhu

Nature Photonics 11, 70 (2017). doi:10.1038/nphoton.2017.1

Authors: Jeffrey M. Gordon & Hui Tong Chua

Nature Photonics 11, 69 (2017). doi:10.1038/nphoton.2016.273

Authors: Marcus Fantham & Clemens F. Kaminski

Dermatologist-level classification of skin cancer with deep neural networks

Nature 542, 7639 (2017). doi:10.1038/nature21056

Authors: Andre Esteva, Brett Kuprel, Roberto A. Novoa, Justin Ko, Susan M. Swetter, Helen M. Blau & Sebastian Thrun

Skin cancer, the most common human malignancy, is primarily diagnosed visually, beginning with an initial clinical screening and followed potentially by dermoscopic analysis, a biopsy and histopathological examination. Automated classification of skin lesions using images is a challenging task owing to the fine-grained variability in the appearance of skin lesions. Deep convolutional neural networks (CNNs) show potential for general and highly variable tasks across many fine-grained object categories. Here we demonstrate classification of skin lesions using a single CNN, trained end-to-end from images directly, using only pixels and disease labels as inputs. We train a CNN using a dataset of 129,450 clinical images—two orders of magnitude larger than previous datasets—consisting of 2,032 different diseases. We test its performance against 21 board-certified dermatologists on biopsy-proven clinical images with two critical binary classification use cases: keratinocyte carcinomas versus benign seborrheic keratoses; and malignant melanomas versus benign nevi. The first case represents the identification of the most common cancers, the second represents the identification of the deadliest skin cancer. The CNN achieves performance on par with all tested experts across both tasks, demonstrating an artificial intelligence capable of classifying skin cancer with a level of competence comparable to dermatologists. Outfitted with deep neural networks, mobile devices can potentially extend the reach of dermatologists outside of the clinic. It is projected that 6.3 billion smartphone subscriptions will exist by the year 2021 (ref. 13) and can therefore potentially provide low-cost universal access to vital diagnostic care.

Implementing the modal method in the electromagnetic grating diffraction problem delivered by the curvilinear coordinate transformation yields a general analytical solution to the 1D grating diffraction problem in a form of a T-matrix. Simultaneously it is shown that the validity of the Rayleigh expansion is defined by the validity of the modal expansion in a transformed medium delivered by the coordinate transformation.

Scientific computing: Code alert

Nature 541, 7638 (2017). doi:10.1038/nj7638-563a

Author: Monya Baker

Programming tools can speed up and strengthen analyses, but mastering the skills takes time and can be daunting.

Author(s): Dorian Bouchet, Mathieu Mivelle, Julien Proust, Bruno Gallas, Igor Ozerov, Maria F. Garcia-Parajo, Angelo Gulinatti, Ivan Rech, Yannick De Wilde, Nicolas Bonod, Valentina Krachmalnicoff, and Sébastien Bidault

Substituting noble metals for high-index dielectrics has recently been proposed as an alternative strategy in nanophotonics, to design broadband optical resonators and circumvent the ohmic losses of plasmonic materials. On the other hand, the authors show that subwavelength silicon nanoantennas can either enhance or inhibit spontaneous emission from fluorescent molecules, a process that is inaccessible with noble metals at the nanoscale. This study highlights the potential of dielectric resonators for low-loss near-field manipulation of solid-state emitters, at room temperature.

[Phys. Rev. Applied 6, 064016] Published Wed Dec 28, 2016

The general features of the light scattering resulting in the so-called resonant Wood's anomalies in the reflection and transmission spectra are described using the effective parameters of a quasi-guided mode. The expression determining the spectral angular dependence of Wood's anomaly in the case of plasmonic grating structures is given and compared to the analogous expression for dielectric grating structures. Comparison of the resonant Wood's anomalies (RWA) with the surface plasmon-polariton resonances (SPPRs) is discussed as well.

Approximating the frequency dispersion of the permittivity of materials with simple analytical functions is of fundamental importance for understanding and modeling the optical response of materials and resulting structures. In the generalized Drude-Lorentz model, the permittivity is described in the complex frequency plane by a number of simple poles having complex weights, which is a physically relevant and mathematically simple approach: By construction, it respects causality represents physical resonances of the material, and can be implemented easily in numerical simulations. We report here an efficient method of optimizing the fit of measured data with the Drude-Lorentz model having an arbitrary number of poles. We show examples of such optimizations for gold, silver, and copper, for different frequency ranges and up to four pairs of Lorentz poles taken into account. We also provide a program implementing the method for general use.

Spinal muscular atrophy, the leading inherited killer of children, has forced generations of parents to watch their kids become progressively weaker and, in severe cases, die by about their second birthdays. The recessively inherited disease inexorably destroys the motor neurons of the spinal cord and brain stem, hobbling muscle movement, including breathing. But now, building on a deep understanding of the biology of the disease, biologists have produced a breakthrough drug that is on the brink of regulatory approval. Nusinersen is an antisense therapy that restores cells' ability to make a protein that is vital to the survival of motor neurons and that is largely absent in children with the disease, who carry a mutated form of the key gene that produces this protein. Two recent, late-stage clinical trials were stopped because nusinersen, developed by Ionis of Carlsbad, California, and now licensed to Biogen of Cambridge, Massachusetts, was so effective in restoring motor function in children and babies that it was deemed unethical to deny the drug to children in the placebo arms of the trials. Neuroscientists are optimistic that the success of nusinersen may portend similar positive results for antisense therapies now being developed to fight other genetically linked nervous system diseases like amyotrophic lateral sclerosis and Huntington disease. Author: Meredith Wadman

Plasmonic nanostructures with enhanced localized optical fields as well as narrow linewidths have driven advances in numerous applications. However, the active engineering of ultranarrow resonances across the visible regime—and within a single system—has not yet been demonstrated. This paper describes how aluminum nanoparticle arrays embedded in an elastomeric slab may...

Simultaneously achieving high optical transparency and excellent charge mobility in semiconducting polymers has presented a challenge for the application of these materials in future “flexible” and “transparent” electronics (FTEs). Here, by blending only a small amount (∼15 wt %) of a diketopyrrolopyrrole-based semiconducting polymer (DPP2T) into an inert polystyrene (PS)...

Magnetic fields from neuronal action potentials (APs) pass largely unperturbed through biological tissue, allowing magnetic measurements of AP dynamics to be performed extracellularly or even outside intact organisms. To date, however, magnetic techniques for sensing neuronal activity have either operated at the macroscale with coarse spatial and/or temporal resolution—e.g., magnetic...

The subwavelength mode volumes of plasmonic filters are well matched to the small size of state-of-the-art active pixels (~ 1 {\mu}m) in CMOS image sensor arrays used in portable electronic devices. Typical plasmonic filters exhibit broad (> 100 nm) transmission bandwidths. Dramatically reducing the peak width of filter transmission spectra would allow for the realization of CMOS hyperspectral imaging arrays, which demand the FWHM of transmission peaks to be less than 30 nm. We find that the design of 5 layer metal-insulator-metal-insulator-metal structures gives rise to multi-mode interference phenomena that suppresses spurious transmission features gives rise to a single narrow transmission band with FWHM as small as 17 nm. The transmission peaks of these multilayer slot-mode plasmonic filters (MSPFs) can be systematically varied throughout the visible and near infrared spectrum, so the same basic structure can serve as a filter over a large range of wavelengths.

An emitter in the vicinity of a metal nanostructure is quenched by its decay through non-radiative channels, leading to the belief in a zone of inactivity for emitters placed within $<$10nm of a plasmonic nanostructure. Here we demonstrate that in tightly-coupled plasmonic resonators forming nanocavities "quenching is quenched" due to plasmon mixing. Unlike isolated nanoparticles, plasmonic nanocavities show mode hybridization which massively enhances emitter excitation and decay via radiative channels. This creates ideal conditions for realizing single-molecule strong-coupling with plasmons, evident in dynamic Rabi-oscillations and experimentally confirmed by laterally dependent emitter placement through DNA-origami.

In this paper, we describe the implementation of leakage radiation microscopy (LRM) to probe the chirality of plasmonic nanostructures. We demonstrate experimentally spin-driven directional coupling as well as vortex generation of surface plasmon polaritons (SPPs) by nanostructures built with T-shaped and $\Lambda$- shaped apertures. Using this far-field method, quantitative inspections, including directivity and extinction ratio measurements, are achieved via polarization analysis in both image and Fourier planes. To support our experimental findings, we develop an analytical model based on a multidipolar representation of $\Lambda$- and T-shaped aperture plasmonic coupler allowing a theoretical explanation of both directionality and singular SPP formation. Furthermore, the roles of symmetry breaking and phases are emphasized in this work. This quantitative characterization of spin-orbit interactions paves the way for developing new directional couplers for subwavelength routing.