Search

Objective To establish a simple and rapid paper-based extraction method for human papilloma virus (HPV) DNA and to evaluate the method. Methods The plasmid containing HPV type 16 gene fragment was employed as the test subjects, and then 6 different paper materials were compared and selected by ultraviolet spectrophotometry and real-time fluorescence quantitative PCR (qPCR). Then, according to the efficiency of nucleic acid extraction and the optimal binding area, a convenient paper strip was designed. Finally, the designed paper strip was applied to extract HPV DNA from 18 clinical cervical exfoliated cell samples (13 positives and 5 negatives), with its performance being evaluated after comparing with the conventional boiling method. Results The qualitative filter paper with a nucleic acid binding zone of 2 mm×10 mm could successfully extract HPV DNA from clinical cervical exfoliated cells. When this test strip was used for DNA extraction, all the 13 high-risk HPV positives were successfully detected out of the 18 clinical samples, which was consistent with the results of the conventional boiling method. In addition, the extraction time of the test strip was less than 1 min for each sample, with a detection limit of 8×102 copies/mL and relative standard deviation of 1.15%. Conclusion Our paper-based rapid extraction for HPV DNA does not require pipetting, heating and centrifugation, and the eluent can be directly used as the PCR template for amplification, which greatly simplifies the extraction process of HPV DNA from clinical cervical exfoliated cell samples.

Compared to human vision, computer vision based on convolutional neural networks (CNN) are more vulnerable to adversarial noises. This difference is likely attributable to how the eyes sample visual input and how the brain processes retinal samples through its dorsal and ventral visual pathways, which are under-explored for computer vision. Inspired by the brain, we design recurrent neural networks, including an input sampler that mimics the human retina, a dorsal network that guides where to look next, and a ventral network that represents the retinal samples. Taking these modules together, the models learn to take multiple glances at an image, attend to a salient part at each glance, and accumulate the representation over time to recognize the image. We test such models for their robustness against a varying level of adversarial noises with a special focus on the effect of different input sampling strategies. Our findings suggest that retinal foveation and sampling renders a model more robust against adversarial noises, and the model may correct itself from an attack when it is given a longer time to take more glances at an image. In conclusion, robust visual recognition can benefit from the combined use of three brain-inspired mechanisms: retinal transformation, attention guided eye movement, and recurrent processing, as opposed to feedforward-only CNNs.

Low detection accuracy and narrow detection range are observed when detecting weak guided wave signals with the Duffing oscillator as a result of a limited frequency band. In order to solve these problems, we proposed a detection method for detecting weak ultrasonic guided waves of arbitrary frequency based on a defect detection method using the Duffing oscillator scanned by a moving window and combined with variable scale and resampling. In this paper, the proposed method by examining the defect echo signals collected on steel pipe samples and detecting ultrasonic guided wave signals excited by arbitrary frequency was effectively validated. Experiment results show that the proposed approach can be used to detect guided wave signals of arbitrary frequency and arbitrary sampling step size with the fixed system. It is very important for expanding the range of defect detection in pipes with guided ultrasonic waves and improving the sensitivity for detecting small defects in engineering applications.

The detection status and existing problems of the anti-soft and heavy object impact performance of glass and metal guardrails for buildings were briefly described in the article. A set of detection equipment was designed for the shortcomings, which mainly included guardrail installation devices, height positioning devices, and impact control systems. It has the advantages of small size, low cost, simple operation, high efficiency and low uncertainty, etc. In addition, the influence of factors such as installation quality and test environment on the test results can be excluded, which greatly improves the accuracy and fairness of the test results.

Vagus nerve stimulation (VNS) is used clinically to treat epilepsy and depression, but its mechanism of action is unknown. Useful techniques to study this are functional magnetic resonance imaging (fMRI) and the local field potential (LFP). fMRI relies on oxygen use in the brain to show areas where neurons are active. The LFP is an electrical signal created by neuron action potentials and other current moving across cell membranes. The most information can be gained when the two methods are used simultaneously, however, this is difficult to do. This study seeks to validate the technique of fMRI-LFP as applied to study the mechanism of VNS. The rat is used as an animal model. Previously collected data is analyzed to determine effects of stimulation on respiration, since this will affect oxygen levels in the blood. Recording electrodes of different materials are tested to find the artifact size created in an MRI environment. Iridium electrodes were found to have the smallest artifact and therefore the best performance. It is unclear whether the stimulation used affects respiration, so a simultaneous fMRI-LFP experiment is needed to interpret fMR images. More work needs to be done before fMRI-LFP recordings can be taken during VNS.

Inspired by "predictive coding" - a theory in neuroscience, we develop a bi-directional and dynamic neural network with local recurrent processing, namely predictive coding network (PCN). Unlike feedforward-only convolutional neural networks, PCN includes both feedback connections, which carry top-down predictions, and feedforward connections, which carry bottom-up errors of prediction. Feedback and feedforward connections enable adjacent layers to interact locally and recurrently to refine representations towards minimization of layer-wise prediction errors. When unfolded over time, the recurrent processing gives rise to an increasingly deeper hierarchy of non-linear transformation, allowing a shallow network to dynamically extend itself into an arbitrarily deep network. We train and test PCN for image classification with SVHN, CIFAR and ImageNet datasets. Despite notably fewer layers and parameters, PCN achieves competitive performance compared to classical and state-of-the-art models. Further analysis shows that the internal representations in PCN converge over time and yield increasingly better accuracy in object recognition. Errors of top-down prediction also reveal visual saliency or bottom-up attention., 12 pages, 3 figures

Dehydrated vocal folds are inefficient sound generators. Although systemic dehydration of the body is believed to induce vocal fold dehydration, this causative relationship has not been demonstrated in vivo. Here we investigate the feasibility of using in vivo proton density (PD)-weighted magnetic resonance imaging (MRI) to demonstrate hydration changes in vocal fold tissue following systemic dehydration in rats.Animal study.Sprague-Dawley rats (n = 10) were imaged at baseline and following a 10% reduction in body weight secondary to withholding water. In vivo, high-field (7 T), PD-weighted MRI was used to successfully resolve vocal fold and salivary gland tissue structures.Normalized signal intensities within the vocal fold decreased postdehydration by an average of 11.38% ± 3.95% (mean ± standard error of the mean [SEM], P = .0098) as compared to predehydration levels. The salivary glands experienced a similar decrease in normalized signal intensity by an average of 10.74% ± 4.14% (mean ± SEM, P = .0195) following dehydration. The correlation coefficient (percent change from dehydration) between vocal folds and salivary glands was 0.7145 (P = .0202).Ten percent systemic dehydration induced vocal fold dehydration as assessed by PD-weighted MRI. Changes in the hydration state of vocal fold tissue were highly correlated with that of the salivary glands in dehydrated rats in vivo. These preliminary findings demonstrate the feasibility of using PD-weighted MRI to quantify hydration states of the vocal folds and lay the foundation for further studies that explore more routine and realistic magnitudes of systemic dehydration and rehydration.NA. Laryngoscope, 128:E222-E227, 2018.

Functional magnetic resonance imaging (fMRI) and electroencephalography (EEG) are useful tools to analyze brain activities given active stimulation. However, the electromagnetic noise from the MRI distorts the brain signal recording and damages the subject with excessive heat generated on the electrodes attached to the skin. MRI-compatible recording and stimulation systems previously developed at LIBI lab were capable of removing the electromagnetic noise during the imaging process. Previously, the hardware systems had required the integrative software that could control both circuits simultaneously and enable users to easily change recording and stimulation parameters. Graphical user interface (GUI) programmed with computer language informed the user to setup safe recording and stimulating parameters and controlled microcontrollers on the MRI-compatible recorder and stimulator. With the stable data transmission, the GUI could send 2 bytes of command and data to a microcontroller on the MRI-compatible stimulation circuit to setup stimulation parameters, inform user of the safety limit, and change desired parameters. Also, the GUI could receive 8-bit resolute recorded data from a microcontroller on the recording circuit and visualize the recorded data from the serial port on the plotting area on the GUI. Therefore, the GUI was successfully integrated into the LIBILAB system to improve safely and provide fast control over the MRI-compatible hardware systems. Furthermore, the GUI will be developed to automatically detect irregular brain response to any change in stimulation parameters and record data with higher resolution.

Concurrent electroencephalogram (EEG) and functional magnetic resonance imaging (fMRI) recordings are susceptible to large amounts of noise due to the static and dynamic magnetic fields present inside the MR scanner. EEG-fMRI studies are conducted to provide better spatial and temporal resolution for each recording, respectively, but the artifacts found in the EEG render the data impossible to interpret. Past studies have focused on signal post-processing techniques which are able to effectively remove noise upon the completion of a study, but there are no techniques able to process the data in real-time without extensive calibration. This research addresses this issue by exploring multi-channel analysis techniques on data combined from each of the EEG channels with a limited number of epochs. In particular, the spatial dependency between the three orthogonal components of the magnetic field gradients was studied to find corresponding templates that can be used to remove this artifact. The resulting algorithms were then tested against an existing algorithm to evaluate their performance. The multi-channel average artifact regression algorithm was found to be robust against the motion artifact and functioned well with few epochs. This algorithm has been implemented in a MATLAB toolbox which will soon be available to the public. Further studies are needed to extend this approach to remove the pulse artifact.

The gear sleeve is the key part of automobile and motorcycle. The top cavity is the small module involute internal gear with addendum modification and many teeth, the gear form is narrow, the tooth tip is pointed. The bottom is taper hole with three notchs, and the concentricity between the top and bottom is high. So, mechanical cutting of the sleeve is difficult, and plastic forming is difficult too. To meet the needs of markets and improving comprehensive function of automobile and motorcycle, the high efficiency cold precise forming technology of net-shaping for the gear sleeve is put forward based on analysis, study and repeated test. In the same time, a set of high accuracy die with simple novel structure and long service life is designed and manufactured. The advantage of the method has been verified that material utilization and production efficiency is high. For this reason, the technology of the cold precise forming is profitable reference for producing of the same type gear sleeve.

Based on the predictive coding theory in neuroscience, we designed a bi-directional and recurrent neural net, namely deep predictive coding networks (PCN). It has feedforward, feedback, and recurrent connections. Feedback connections from a higher layer carry the prediction of its lower-layer representation; feedforward connections carry the prediction errors to its higher-layer. Given image input, PCN runs recursive cycles of bottom-up and top-down computation to update its internal representations and reduce the difference between bottom-up input and top-down prediction at every layer. After multiple cycles of recursive updating, the representation is used for image classification. With benchmark data (CIFAR-10/100, SVHN, and MNIST), PCN was found to always outperform its feedforward-only counterpart: a model without any mechanism for recurrent dynamics. Its performance tended to improve given more cycles of computation over time. In short, PCN reuses a single architecture to recursively run bottom-up and top-down processes. As a dynamical system, PCN can be unfolded to a feedforward model that becomes deeper and deeper over time, while refining it representation towards more accurate and definitive object recognition., Comment: 10 pages, 5 figures, 4 tables