Artificial convolution neural network techniques and applications for lung nodule detection

We have developed a double-matching method and an artificial visual neural network technique for lung nodule detection. This neural network technique is generally applicable to the recognition of medical image pattern in gray scale imaging. The structure of the artificial neural net is a simplified network structure of human vision. The fundamental operation of the artificial neural network is local two-dimensional convolution rather than full connection with weighted multiplication. Weighting coefficients of the convolution kernels are formed by the neural network through backpropagated training. In addition, we modeled radiologists' reading procedures in order to instruct the artificial neural network to recognize the image patterns predefined and those of interest to experts in radiology. We have tested this method for lung nodule detection. The performance studies have shown the potential use of this technique in a clinical setting. This program first performed an initial nodule search with high sensitivity in detecting round objects using a sphere template double-matching technique. The artificial convolution neural network acted as a final classifier to determine whether the suspected image block contains a lung nodule. The total processing time for the automatic detection of lung nodules using both prescan and convolution neural network evaluation was about 15 seconds in a DEC Alpha workstation.     

Fabrication of Ordered Nanostructured Arrays Using Poly(dimethylsiloxane) Replica Molds Based on Three‐Dimensional Colloidal Crystals

Hexagonally arrayed structures of colloidal crystals with uniform surface are a good candidate for master molds to be used in soft lithography. Here, the fabrication of periodically arrayed nanostructures using poly(dimethylsiloxane) (PDMS) molds based on three‐dimensionally (3D) ordered colloidal crystals is reported. A robust, high‐quality 3D colloidal‐crystal master molds is prepared using the colloidal suspension containing a water‐soluble polymer. The surface patterns of the 3D colloidal crystals can then be transferred onto a polymer film via soft lithography, by means of the replication of the surface pattern with PDMS. Various hexagonally arrayed nanostructure patterns can be fabricated, including close‐packed and non‐close‐packed 2D arrays and honeycomb structures by the structural modification of the 3D colloidal‐crystal templates. The replicated hexagonally arrayed structures can also be used as templates for producing colloidal crystals with 2D superlattices.     

Performance of 2-D RAKE receiver in a correlatedfrequency-selective Nakagami-fading

The bit error rate (BER) performance and the characteristics of a two-dimensional (2-D) RAKE receiver operating in a correlated frequency-selective Nakagami-fading environment are analyzed. Correlated fading between array elements whose fading statistics are identical across the same RAKE branch, as well as an arbitrary number of RAKE-branches with arbitrary finding statistics, are assumed. We derived an approximated signal-to-noise ratio (SNR) statistics for one RAKE branch with correlated multiple antennas, which is extended to that for multiple RAKE branches with arbitrary fading statistics, i.e., a 2-D RAKE receiver. The receiver's performance and characteristics are analyzed using the cumulative distribution function of the SNR at the 2-D RAKE receiver output and the BER under various conditions, Numerical results show that the improvement In performance of the 2-D RAKE receiver is brought about by the average SNR and diversity gains, which are identified by two parameters specifying the gamma distribution of SNR     

Accurate modeling of the effects of fringing area interface traps on scanning capacitance microscopy measurement

Scanning capacitance microscopy (SCM) is a dopant profile extraction tool with nanometer spatial resolution. While it is based on the high-frequency MOS capacitor theory, there are crucial second-order effects which make the extraction of dopant profile from SCM data a challenging task. Due to the small size of the SCM probe, the trapped charges in the interface traps at the oxide-silicon dioxide interface surrounding the probe significantly affect the measured SCM data through the fringing electric field created by the trapped charges. In this paper, we present numerical simulation results to investigate the nature of SCM dC/dV data in the presence of interface traps. The simulation takes into consideration the traps' response to the ac signal used to measure dC/dV as well as the fringing field of the trapped charge surrounding the probe tip. In this paper, we present an error estimation of experimental SCM dopant concentration extraction when the interface traps and fringing field are ignored. The trap distribution in a typical SCM sample is also investigated.     

Self‐Assembly of Nanoparticle‐Spiked Pillar Arrays for Plasmonic Biosensing

Plasmonic biosensors have demonstrated superior performance in detecting various biomolecules with high sensitivity through simple assays. Scaled‐up, reproducible chip production with a high density of hotspots in a large area has been technically challenging, limiting the commercialization and clinical translation of these biosensors. A new fabrication method for 3D plasmonic nanostructures with a high density, large volume of hotspots and therefore inherently improved detection capabilities is developed. Specifically, Au nanoparticle‐spiked Au nanopillar arrays are prepared by utilizing enhanced surface diffusion of adsorbed Au atoms on a slippery Au nanopillar arrays through a simple vacuum process. This process enables the direct formation of a high density of spherical Au nanoparticles on the 1 nm‐thick dielectric coated Au nanopillar arrays without high‐temperature annealing, which results in multiple plasmonic coupling, and thereby large effective volume of hotspots in 3D spaces. The plasmonic nanostructures show signal enhancements over 8.3 × 10⁸‐fold for surface‐enhanced Raman spectroscopy and over 2.7 × 10²‐fold for plasmon‐enhanced fluorescence. The 3D plasmonic chip is used to detect avian influenza‐associated antibodies at 100 times higher sensitivity compared with unstructured Au substrates for plasmon‐enhanced fluorescence detection. Such a simple and scalable fabrication of highly sensitive 3D plasmonic nanostructures provides new opportunities to broaden plasmon‐enhanced sensing applications.     

Sub‐Nanometer Level Size Tuning of a Monodisperse Nanoparticle Array Via Block Copolymer Lithography

The fabrication and catalytic application of a size‐tunable monodisperse nanoparticle array enabled by block copolymer lithography is demonstrated. Highly uniform vertical cylinder nanodomains are achieved in poly(styrene‐block‐4‐vinylpyridine) (PS‐b‐P4VP) diblock copolymer thin‐films by solvent annealing. The prominent diffusion of the anionic metal complexes into the protonated P4VP cylinder nanodomains occurs through specific electrostatic interactions in a weakly acidic aqueous solution. This well‐defined diffusion with nanoscale confinement enables preparation of the laterally ordered monodisperse nanoparticle array with sub‐nanometer level precise size tuning. The controlled growth of monodisperse nanoparticle arrays is proven by their catalytic use for vertical carbon nanotube (CNT) growth via plasma enhanced chemical vapor deposition (PECVD). Since the size of the catalyst particles is the decisive parameter for the diameters and wall‐numbers of CNTs, the highly selective growth of double‐walled or triple‐walled CNTs could be accomplished using monodisperse nanoparticle arrays.     

Biomimetic Chitin–Silk Hybrids: An Optically Transparent Structural Platform for Wearable Devices and Advanced Electronics

The cuticles of insects and marine crustaceans are fascinating models for man‐made advanced functional composites. The excellent mechanical properties of these biological structures rest on the exquisite self‐assembly of natural ingredients, such as biominerals, polysaccharides, and proteins. Among them, the two commonly found building blocks in the model biocomposites are chitin nanofibers and silk‐like proteins with β‐sheet structure. Despite being wholly organic, the chitinous protein complex plays a key role for the biocomposites by contributing to the overall mechanical robustness and structural integrity. Moreover, the chitinous protein complex alone without biominerals is optically transparent (e.g., dragonfly wings), thereby making it a brilliant model material system for engineering applications where optical transparency is essentially required. Here, inspired by the chitinous protein complex of arthropods cuticles, an optically transparent biomimetic composite that hybridizes chitin nanofibers and silk fibroin (β‐sheet) is introduced, and its potential as a biocompatible structural platform for emerging wearable devices (e.g., smart contact lenses) and advanced displays (e.g., transparent plastic cover window) is demonstrated.     

Trenched‐Sinker LDMOSFET (TS‐LDMOS) Structure for 2 GHz Power Amplifiers

This paper proposes a new LDMOSFET structure with a trenched sinker for high‐power RF amplifiers. Using a low‐temperature, deep‐trench technology, we succeeded in drastically shrinking the sinker area to one‐third the size of the conventional diffusion‐type structure. The RF performance of the proposed device with a channel width of 5 mm showed a small signal gain of 16.5 dB and a maximum peak power of 32 dBm with a power‐added efficiency of 25% at 2 GHz. Furthermore, the trench sinker, which was applied to the guard ring to suppress coupling between inductors, showed an excellent blocking performance below ?40 dB at a frequency of up to 20 GHz. These results confirm that the proposed trenched sinker should be an effective technology both as a compact sinker for RF power devices and as a guard ring against coupling.     

Design of Parasitic Inductance Reduction in GaN Cascode FET for High‐Efficiency Operation

This paper presents a method of parasitic inductance reduction for high‐speed switching and high‐efficiency operation of a cascode structure with a low‐voltage enhancement‐mode silicon (Si) metal–oxide–semiconductor field‐effect transistor (MOSFET) and a high‐voltage depletion‐mode gallium nitride (GaN) field‐effect transistor (FET). The method is proposed to add a bonding wire interconnected between the source electrode of the Si MOSFET and the gate electrode of the GaN FET in a conventional cascode structure package to reduce the most critical inductance, which provides the major switching loss for a high switching speed and high efficiency. From the measured results of the proposed and conventional GaN cascode FETs, the rising and falling times of the proposed GaN cascode FET were up to 3.4% and 8.0% faster than those of the conventional GaN cascode FET, respectively, under measurement conditions of 30 V and 5 A. During the rising and falling times, the energy losses of the proposed GaN cascode FET were up to 0.3% and 6.7% lower than those of the conventional GaN cascode FET, respectively.