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.

Subwavelength pixelated CMOS color sensors based on anti-Hermitian metasurface.

Abstract: The demand for essential pixel components with ever-decreasing size and enhanced performance is central to current optoelectronic applications, including imaging, sensing, photovoltaics and communications. The size of the pixels, however, are severely limited by the fundamental constraints of lightwave diffraction. Current development using transmissive filters and planar absorbing layers can shrink the pixel size, yet there are two major issues, optical and electrical crosstalk, that need to be addressed when the pixel dimension approaches wavelength scale. All these fundamental constraints preclude the continual reduction of pixel dimensions and enhanced performance. Here we demonstrate subwavelength scale color pixels in a CMOS compatible platform based on anti-Hermitian metasurfaces. In stark contrast to conventional pixels, spectral filtering is achieved through structural color rather than transmissive filters leading to simultaneously high color purity and quantum efficiency. As a result, this subwavelength anti-Hermitian metasurface sensor, over 28,000 pixels, is able to sort three colors over a 100 nm bandwidth in the visible regime, independently of the polarization of normally-incident light. Furthermore, the quantum yield approaches that of commercial silicon photodiodes, with a responsivity exceeding 0.25 A/W for each channel. Our demonstration opens a new door to sub-wavelength pixelated CMOS sensors and promises future high-performance optoelectronic systems. Pixel size in imaging and displays is limited by fundamental constraints that compromise performance at wavelength scales. Here the authors present subwavelength color pixel sensors based on anti-Hermitian metasurfaces relying on structural color for increased performance.

Harnessing the power of antibodies to fight bone metastasis

Abstract: Antibody-based therapies have proved to be of great value in cancer treatment. Despite the clinical success of these biopharmaceuticals, reaching targets in the bone microenvironment has proved to be difficult due to the relatively low vascularization of bone tissue and the presence of physical barriers. Here, we have used an innovative bone-targeting (BonTarg) technology to generate a first-in-class bone-targeting antibody. Our strategy involves the use of pClick antibody conjugation technology to chemically couple the bone-targeting moiety bisphosphonate to therapeutic antibodies. Bisphosphonate modification of these antibodies results in the delivery of higher conjugate concentrations to the bone metastatic niche, relative to other tissues. In xenograft mice models, this strategy provides enhanced inhibition of bone metastases and multiorgan secondary metastases that arise from bone lesions. Specific delivery of therapeutic antibodies to the bone, therefore, represents a promising strategy for the treatment of bone metastatic cancers and other bone diseases.

circCUL2 induces an inflammatory CAF phenotype in pancreatic ductal adenocarcinoma via the activation of the MyD88-dependent NF-κB signaling pathway

Abstract: Pancreatic ductal adenocarcinoma (PDAC) is characterized by clusters of cancer cells surrounded by a dense desmoplastic stroma. However, little is known about stromal cell heterogeneity in the pancreatic tumor microenvironment.We conducted circRNA profiling in primary fibroblasts by high-throughput sequencing and detected circCUL2 levels in PDAC tissues by qRT-PCR. We subsequently investigated the effect of circCUL2 on inflammatory cancer-associated fibroblast (iCAF) activation, heterogeneity and protumor activity by ELISA, flow cytometry, colony formation and transwell assays in vitro and by xenograft models in vivo. The regulatory effect of circCUL2 on miR-203a-3p/MyD88/IL6 was examined by RNA pulldown, FISH, and luciferase reporter assays.We identified that circCUL2 was specifically expressed in cancer-associated fibroblasts (CAFs) but not in cancer cells. Moreover, the enrichment of circCUL2 in tumor tissues was significantly correlated with the poor prognosis of PDAC patients. Upregulation of circCUL2 expression in normal fibroblasts (NFs) induced the iCAF phenotype, and then iCAFs promoted PDAC progression through IL6 secretion in vitro. Furthermore, circCUL2-transduced NFs promoted tumorigenesis and metastasis of PDAC cells in vivo, which was blocked by an anti-IL6 antibody. Mechanistically, circCUL2 functioned as a ceRNA and modulated the miR-203a-3p/MyD88/NF-κB/IL6 axis, thereby further activating the STAT3 signaling pathway in pancreatic cancer cells to induce PDAC progression.We showed that the circCUL2/miR-203a-5p/MyD88/NF-κB/IL6 axis contributes to the induction of iCAFs and established a distinct fibroblast niche for PDAC progression, which could help the development of strategies that selectively target tumor-promoting CAFs in PDAC.

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.