Time filter

Source Type

Menon P.G.,Carnegie Mellon University | Menon P.G.,Shunde International Joint Research Institute | Menon P.G.,QuantMD LLC | Muller J.H.,Stellenbosch University
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2014

We introduce patellofemoral separation (PFS) as a novel metric to quantify patella-trochlear proximity as a function of dynamic knee flexion. PFS is quantified in 4D (i.e. 3D+time) using accurate segmentation from pre-operative imaging data acquired in three discrete, quasi-static knee postures, up to the maximum bending limit (i.e. 40° of flexion), within the constraints of a standard computed tomography (CT) or magnetic resonance imaging (MRI) scanner. Additionally, in this study, in order to examine patient-specific patella postures over a full range from 0 to 90° of dynamic knee flexion and extension, we utilize a computational model to simulate dynamic patella kinematics beyond 40° of bending. The computational model was optimized to reproduce patella postures as determined from the imaging data. A method of shape-based interpolation of the acquired 3D components (i.e. bone and cartilage) of the knee was applied in order to recreate a continuous range of motion of the patella and femur during knee bending from 0° to 40° using imaging data and 0° to 90° from simulated data. Next, a regional Hausdorff distance mapping paradigm was applied to compare the separation of the 3D surfaces defined by the patella and femoral cartilage segmentations from the interpolated imaging-based and simulated knee postures, at 1°increments. This separation distance was termed as PFS and examined as a posture-varying color map on the patella cartilage surface. The mean PFS was computed as the mean HD of separation between patella and femoral cartilage, at each posture over the entire studied range of motion. Mean PFS was observed to decrease with increased knee flexion, evidencing increased proximity of the patella and femur and increased risk of contact. In order to automatically quantify signs of patellofemoral instability from pathological knee kinematics reconstructed using medical imaging, the limits of PFS defining the thresholds of pain will require to be determined by benchmarking the metric against patients with normal knee-function. The PFS metric may also find potential application as a biomarker for the identification of high localized patellofemoral pressure by predicting patellofemoral impingement. © 2014 Springer International Publishing.

Long D.X.,Dongguk University | Xu Y.,Dongguk University | Wei H.-X.,Dongguk University | Liu C.,Sun Yat Sen University | And 2 more authors.
Physical Chemistry Chemical Physics | Year: 2015

A simply and facilely synthesized MoO3 solution was developed to fabricate charge injection layers for improving the charge-injection properties in p-type organic field-effect transistors (OFETs). By dissolving MoO3 powder in ammonium (NH3) solvent under an air atmosphere, an intermediate ammonium molybdate ((NH4)2MoO4) precursor is made stable, transparent and spin-coated to form the MoO3 interfacial layers, the thickness and morphology of which can be well-controlled. When the MoO3 layer was applied to OFETs with a cost-effective molybdenum (Mo) electrode, the field-effect mobility (μFET) was significantly improved to 0.17 or 1.85 cm2 V-1s-1 for polymer semiconductors, regioregular poly(3-hexylthiophene) (P3HT) or 3,6-bis-(5bromo-thiophen-2-yl)-N,N′-bis(2-octyl-1-dodecyl)-1,4-dioxo-pyrrolo[3,4-c]pyrrole (DPPT-TT), respectively. Device analysis indicates that the MoO3-deposited Mo contact exhibits a contact resistance RC of 1.2 MΩ cm comparable to that in a device with the noble Au electrode. Kelvin-probe measurements show that the work function of the Mo electrode did not exhibit a dependence on the thickness of MoO3 film. Instead, ultraviolet photoemission spectroscopy results show that a doping effect is probably induced by casting the MoO3 layer on the P3HT semiconductor, which leads to the improved hole injection. © 2015 the Owner Societies.

Cai K.,Sun Yat Sen University | Jiang M.,Sun Yat Sen University | Jiang M.,Shunde International Joint Research Institute
IEEE Photonics Journal | Year: 2016

In this paper, we propose a multiuser precoded visible light communication (VLC) system utilizing optical orthogonal frequency division multiplexing (OFDM), where optical spatial modulation (OSM) and spatial pulse position modulation (SPPM) are invoked for supporting high-rate data transmissions without compromising uniform illumination across multiple light-emitting diodes (LEDs). Both OSM and SPPM exploit the index of data streams for carrying additional information bits, whereas the multiuser interference occurring between different user equipments is mitigated by the MIMO precoder. By evaluating the performances of the proposed VLC systems, we demonstrate that the OSM-aided system is capable of achieving high data rates at sufficiently high signal-to-noise ratios (SNRs). In contrast, the SPPM-aided system has a higher effective bandwidth efficiency at low SNRs, despite the fact that its attainable peak data rate is lower than that of the SM-aided counterpart. Such tradeoffs indicate the potential application scenarios of the proposed schemes for achieving diverse design prerequisites. © 2016 IEEE.

Liu J.,Sun Yat Sen University | Liu J.,Shunde International Joint Research Institute | Zheng S.,Sun Yat Sen University | Zheng S.,Shunde International Joint Research Institute | And 4 more authors.
IEEE Antennas and Wireless Propagation Letters | Year: 2014

A center-fed circular patch antenna with shorting vias and a coupled annular ring is proposed. With a low-profile configuration, the antenna provides a wide bandwidth by merging three resonant modes, including the TM02 mode of the circular patch, the TM01 mode generated by the shorting vias, and the TM02 mode of the coupled annular ring. The antenna operating in these modes would produce an omnidirectional pattern in the horizontal plane similar to that generated by a monopole antenna. A reduced structure is used to simplify the simulation in optimizing the bandwidth of the antenna. Measured results show that the antenna achieves a bandwidth of 27.4% for a profile of 0.029 wavelengths and has a gain of about 6 dBi. © 2014 IEEE.

Liang X.,Sun Yat Sen University | Lin L.,Sun Yat Sen University | Lin L.,Shunde International Joint Research Institute | Cao Q.,Sun Yat Sen University | And 2 more authors.
IEEE Transactions on Medical Imaging | Year: 2016

This work investigates how to automatically classify Focal Liver Lesions (FLLs) into three specific benign or malignant types in Contrast-Enhanced Ultrasound (CEUS) videos, and aims at providing a computational framework to assist clinicians in FLL diagnosis. The main challenge for this task is that FLLs in CEUS videos often show diverse enhancement patterns at different temporal phases. To handle these diverse patterns, we propose a novel structured model, which detects a number of discriminative Regions of Interest (ROIs) for the FLL and recognize the FLL based on these ROIs. Our model incorporates an ensemble of local classifiers in the attempt to identify different enhancement patterns of ROIs, and in particular, we make the model reconfigurable by introducing switch variables to adaptively select appropriate classifiers during inference. We formulate the model learning as a non-convex optimization problem, and present a principled optimization method to solve it in a dynamic manner: the latent structures (e.g. the selections of local classifiers, and the sizes and locations of ROIs) are iteratively determined along with the parameter learning. Given the updated model parameters in each step, the data-driven inference is also proposed to efficiently determine the latent structures by using the sequential pruning and dynamic programming method. In the experiments, we demonstrate superior performances over the state-of-the-art approaches. We also release hundreds of CEUS FLLs videos used to quantitatively evaluate this work, which to the best of our knowledge forms the largest dataset in the literature. Please find more information at "http://vision. sysu.edu.cn/projects/fllrecog/". © 2015 IEEE.

Yi Y.,Sun Yat Sen University | Yi Y.,Shunde International Joint Research Institute | Xu H.,Sun Yat Sen University | Xu H.,Shunde International Joint Research Institute
IEEE Signal Processing Letters | Year: 2014

The problem of tracking multiple targets in video is addressed, and a novel hierarchical data association framework with occlusion handling is presented. The association hierarchy first divides the detections, targets, and candidates into several disjoint branches, then progressively associates the detections to the targets and candidates within each branch, and finally initializes the targets through candidate upgrade. Furthermore, the depth disorder inference and targets motion pairing prediction are introduced to explicitly tackle target-target and target-environment occlusions, respectively. Experiments verify that our method improves the runtime performance significantly while keeping competitive tracking accuracy and precision compared with several state-of-the-art methods. © 2014 IEEE.

Lin L.,Sun Yat Sen University | Wang X.,Shunde International Joint Research Institute | Yang W.,Sun Yat Sen University | Lai J.-H.,Sun Yat Sen University
IEEE Transactions on Pattern Analysis and Machine Intelligence | Year: 2015

In this paper, we investigate a novel reconfigurable part-based model, namely And-Or graph model, to recognize object shapes in images. Our proposed model consists of four layers: leaf-nodes at the bottom are local classifiers for detecting contour fragments; or-nodes above the leaf-nodes function as the switches to activate their child leaf-nodes, making the model reconfigurable during inference; and-nodes in a higher layer capture holistic shape deformations; one root-node on the top, which is also an or-node, activates one of its child and-nodes to deal with large global variations (e.g. different poses and views). We propose a novel structural optimization algorithm to discriminatively train the And-Or model from weakly annotated data. This algorithm iteratively determines the model structures (e.g. the nodes and their layouts) along with the parameter learning. On several challenging datasets, our model demonstrates the effectiveness to perform robust shape-based object detection against background clutter and outperforms the other state-of-the-art approaches. We also release a new shape database with annotations, which includes more than 1500 challenging shape instances, for recognition and detection. © 2014 IEEE.

Zhu J.-Y.,Sun Yat Sen University | Zhu J.-Y.,Shunde International Joint Research Institute | Zheng W.-S.,Sun Yat Sen University | Zheng W.-S.,Guangdong Province Key Laboratory of Computational Science | And 4 more authors.
IEEE Transactions on Information Forensics and Security | Year: 2014

Visual versus near infrared (VIS-NIR) face image matching uses an NIR face image as the probe and conventional VIS face images as enrollment. It takes advantage of the NIR face technology in tackling illumination changes and low-light condition and can cater for more applications where the enrollment is done using VIS face images such as ID card photos. Existing VIS-NIR techniques assume that during classifier learning, the VIS images of each target people have their NIR counterparts. However, since corresponding VIS-NIR image pairs of the same people are not always available, which is often the case, so those methods cannot be applied. To address this problem, we propose a transductive method named transductive heterogeneous face matching (THFM) to adapt the VIS-NIR matching learned from training with available image pairs to all people in the target set. In addition, we propose a simple feature representation for effective VIS-NIR matching, which can be computed in three steps, namely Log-DoG filtering, local encoding, and uniform feature normalization, to reduce heterogeneities between VIS and NIR images. The transduction approach can reduce the domain difference due to heterogeneous data and learn the discriminative model for target people simultaneously. To the best of our knowledge, it is the first attempt to formulate the VIS-NIR matching using transduction to address the generalization problem for matching. Experimental results validate the effectiveness of our proposed method on the heterogeneous face biometric databases. © 2005-2012 IEEE.

Menon P.G.,Sun Yat Sen University | Menon P.G.,Shunde International Joint Research Institute | Menon P.G.,QuantMD LLC | Ludwig D.,University of Pittsburgh | And 3 more authors.
Journal of Interventional Cardiology | Year: 2014

Background Preliminary clinical experience with a percutaneous endoventricular partitioning device (Parachute®, CardioKinetix Inc., Menlo Park, CA, USA) suggests that it ameliorates global LV dysfunction and heart failure symptoms in selected patients who have suffered previous anterior myocardial infarction. Less is known of its effect on regional LV function.Objective To gain insight into device effect on regional LV function by analysis of cardiac computed tomographic (CT) images obtained before and after device implantation.Methods Comparative analysis of pre and 6 months post-implantation contrast-enhanced CT images from 6 subjects enrolled in the phase 1 Parachute clinical trials, including regional LV volume and systolic excursion, as well as device motion.Results After implantation, a significant reduction in volume of the "dynamic" LV compartment (that which was not excluded by the device) was accompanied by a significant reduction in dykinetic motion and a trend toward an improved ejection fraction. Penetration of contrast into the excluded compartment was still present at 6 months, however the apical motion was significantly less diskinetic in 3 subjects and unchanged in the other 3. Overall device surface motion was inward in systole, a significant improvement relative to the overall dyskinetic LV apex pre-implantation. Device motion was spatially heterogeneous, which appeared to be dependent on the motility of the myocardium that anchored its individual splines.Conclusions Our data suggest that the Parachute device acts as a functional impediment to flow and stretch, effectively depressurizing the apical segment. © 2014, Wiley Periodicals, Inc.

Piao D.,Carnegie Mellon University | Menon P.G.,SYSU CMU Joint Institute of Engineering | Menon P.G.,Shunde International Joint Research Institute | Mengshoel O.J.,Carnegie Mellon University
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2014

Optical flow methods are often used in image processing, for example for object recognition and image segmentation. Traditional optical flow methods use numerical methods, assuming intensity constancy of pixels' movements. In this work we describe a probabilistic method of modeling the optical flow problem, and discuss the use of Gibbs sampling for optimization of the computed optical flow vector field. In experiments involving test images as well as medical image slices through the short-axis of the left ventricle of the heart, our probabilistic method is compared with the classic Horn-Schunck optical flow method. We demonstrate that our proposed approach probabilistic optical flow method is robust to changes in the shape and intensity of objects tracked. This is a useful property when identifying cardiac structures from time-resolved medical images of the heart, where the shape of the cardiac structures change between consecutive temporal frames of the cardiac cycle. © 2014 Springer International Publishing.

Loading Shunde International Joint Research Institute collaborators
Loading Shunde International Joint Research Institute collaborators