Xiang Zhang

Bio: Xiang Zhang is an academic researcher from Baylor College of Medicine. The author has contributed to research in topics: Medicine & Computer science. The author has an hindex of 154, co-authored 1733 publications receiving 117576 citations. Previous affiliations of Xiang Zhang include University of California, Berkeley & University of Texas MD Anderson Cancer Center.

Bone-in-culture array as a platform to model early-stage bone metastases and discover anti-metastasis therapies

Abstract: The majority of breast cancer models for drug discovery are based on orthotopic or subcutaneous tumours. Therapeutic responses of metastases, especially microscopic metastases, are likely to differ from these tumours due to distinct cancer-microenvironment crosstalk in distant organs. Here, to recapitulate such differences, we established an ex vivo bone metastasis model, termed bone-in-culture array or BICA, by fragmenting mouse bones preloaded with breast cancer cells via intra-iliac artery injection. Cancer cells in BICA maintain features of in vivo bone micrometastases regarding the microenvironmental niche, gene expression profile, metastatic growth kinetics and therapeutic responses. Through a proof-of-principle drug screening using BICA, we found that danusertib, an inhibitor of the Aurora kinase family, preferentially inhibits bone micrometastases. In contrast, certain histone methyltransferase inhibitors stimulate metastatic outgrowth of indolent cancer cells, specifically in the bone. Thus, BICA can be used to investigate mechanisms involved in bone colonization and to rapidly test drug efficacies on bone micrometastases.

Evaluation of residual stresses induced by cold spraying of Ti-6Al-4V on Ti-6Al-4V substrates

Abstract: Cold spray (CS) is a solid-state additive material deposition technique, which has gained attention in the aerospace industry as a potentially viable technology for structural repair of high-value parts made of high-strength alloys such as Ti-6Al-4V (Ti-64). Residual stresses build up in the substrate and deposited materials resulting from the CS process can influence the integrity of a coating or repair. However, the nature, magnitude and distribution of residual stresses in Ti-64/Ti-64 CS repairs are currently unknown. This study aims to evaluate the effects of geometrical variables (i.e. the number of CS layers, CS layer thickness, and substrate thickness) and track pattern on the magnitude and distribution of residual stresses in CS deposit-substrate assemblies. Through-thickness stress distributions were measured experimentally by neutron diffraction and contour method. Furthermore, a comparison among different residual stress build-up mechanisms induced by CS processes has been discussed for different combinations of substrate and deposit assemblies. An analytical model based on the force and moment equilibrium requirements was used to interpret the experimental stress profiles and to predict the residual stress distribution. It was found that residual stresses are highly tensile near the free surface of the Ti-64 deposits as well as towards the bottom of the substrate, and compressive near the interface region. Although all the specimens showed similar stress distribution, the magnitudes were found to be higher in one or more of the following cases: specimens with a higher number of CS layers, lower substrate thickness, higher layer thickness (i.e. at lower scanning speed), and deposited with a horizontal track pattern.

Valley optomechanics in a monolayer semiconductor

Abstract: Interfacing nanomechanics with photonics and charge/spin-based electronics has transformed information technology and facilitated fundamental searches for the quantum-to-classical transition1–3. Utilizing the electron valley degree of freedom as an information carrier, valleytronics has recently emerged as a promising platform for developments in computation and communication4–7. Thus far, explorations of valleytronics have focused on optoelectronic and magnetic means8–16. Here, we realize valley–mechanical coupling in a resonator made of the monolayer semiconductor MoS2 and transduce valley information into mechanical states. The coupling is achieved by exploiting the magnetic moment of valley carriers with a magnetic field gradient. We optically populate the valleys and observe the resulting mechanical actuation using laser interferometry. We are thus able to control the valley–mechanical interaction by adjusting the pump-laser light, the magnetic field gradient and temperature. Our work paves the way for realizing valley-actuated devices and hybrid valley quantum systems. Transduction of valley information to mechanical states in a monolayer MoS2 resonator can be realized by optically pumping the valley carriers and applying an out-of-plane magnetic field gradient to induce a displacement-dependent valley splitting.

Elastic scattering of the proton drip-line nucleus 8 B off a nat Pb target at 170.3 MeV

Abstract: The elastic scattering of ${}^{8}$B by a ${}^{\mathrm{nat}}$Pb target was measured at an incident energy of 170.3 MeV. Special care was taken with the limited intensity and broad profile of the secondary beam. The measured angular distribution of the differential cross section shows that the Coulomb-nuclear interference peak (CNIP) is not suppressed in this system, in contrast to what was observed in the scattering of neutron halo nuclei by heavy targets at energies around the Coulomb barrier. Analyses of the angular distribution were performed both in terms of the optical model using a single-folding-type potential and the continuum discretized coupled-channels (CDCC) method, which explicitly takes into account the breakup-channel couplings to the elastic scattering. The overall pattern of the differential cross section is well reproduced by the CDCC calculations. The calculations show that the effect of breakup-channel couplings on the elastic scattering is small in the present case.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Going deeper with convolutions

Abstract: We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.