Search

Background Early detection of lung cancer in asymptomatic patients remains challenging, especially for stage I. Considering the substantial interaction with tumor immunogenicity, we hypothesized that lung cancer-associated TCR (LC-aTCR) may serve as potential biomarker in early detection of stage I lung cancer. Methods Individuals who received low-dose computed tomography (LDCT) screening were enrolled in the study. Surgical tissues and peripheral blood specimens were collected and performed with DNA-based T cell repertoire (TCR) sequencing. The motif-based algorithm was used to deconstruct specific lung cancer-associated TCRs (LC-aTCRs). Results A total of 146 individuals participating in the real-world LDCT screening project were enrolled in this study, including 52 patients with pathologically-confirmed stage I lung cancer and 94 non-cancer controls. We developed a motif-based algorithm to define 80 LC-aTCRs in the training cohort. Moreover, in the validation cohort, high sensitivity and specificity was showed in stage I lung cancer with 72% and 91% respectively, and the AUC of the ROC curve was 0.91 (95% CI: 0.85 ∼ 0.96). Conclusion This work provides inspiration for stage I lung cancer detection by using blood TCR profiling data. The combination of TCR-based assay and routine screening deserves further testing in larger cohorts.

With the continuous reform of intelligent manufacturing, industrial production has gradually developed from automation to intelligence. The fusion of vision technology and industrial machines has become a hot research direction in current intelligent transformation. However, machines are not as flexible as humans when grabbing, and still have great limitations. Affected by various characteristics of target objects, such as shape, material, weight and other factors, as well as complex and changeable environmental factors, the research of machine grabbing still faces severe challenges. For the actual complex working conditions, the poor target detection effect leads to the inability to complete accurate grabbing, which affects the production efficiency. This paper proposes a grabbing system with convolutional neural network, which can achieve target detection, classification, positioning and grabbing tasks. First, by comparing the current mainstream target recognition and detection algorithms, select SSD that have both real-time performance and accuracy. Then make specific network structure improvements according to the detection requirements, and insert the Inception structure. At the same time optimize its loss function and nonmaximum suppression. The improved recognition rate is higher, and the target detection frame is closer to the real part, which greatly reduces the recognition error. Second, this research proposes an algorithm model for regional posture detection and grabbing positioning, which uses the output of the previous stage as input to perform posture detection and grabbing positioning of the grabbed target. In the network, the posture angle of the grabbing target is output in a classified manner, and the position coordinates of the grabbing point are output using a regression method. Experiments have proved that our method can perform efficient target recognition and grabbing positioning.

A novel cable driven parallel robot is introduced to assist upper limb in rehabilitation treatments in this paper, which has a lightweight structure. This robot is characterized by a novel and flexible end-effector that is easy to change assembly for different therapies. And it is designed for desktop usage, so its size is not too large. After presenting the structure, the kinematic model is presented for optimal design. Then, the minimal singular value is considered as the velocity transmission index. For upper limb rehabilitation, a new optimal design problem with the transmission index and the robot size is formulated. Last, the optimization presents a design with satisfying performances in velocity transmission capability and robot size.

Objective KRAS mutation plays a critical role in the initiation and development of non‐small cell lung cancer (NSCLC). KRAS‐mutant patients exhibit diverse response to chemotherapy. KRAS co‐mutation subtypes and their prognosis value in advanced Chinese NSCLC patients remain largely elusive. Methods A total of 1126 Chinese advanced NSCLC patients from Xiangya hospital were screened by capture‐based ultra‐deep sequencing for KRAS mutation between January 2015 and December 2016. Survival analyses were performed using Kaplan‐Meier analysis. Results Among the patients screened, 84 cases were detected with KRAS mutation (7.5%). All of them were non‐squamous NSCLC and received pemetrexed plus platinum as the first‐line treatment. The most frequent KRAS co‐mutation genes were TP53 (29%), TP53/LKB1 (19%), and LKB1 (14%). Our data revealed that patients with KRAS co‐mutation had poorer prognosis in comparison with those harboring single KRAS mutation. Furthermore, patients with KPL (KRAS mutated with TP53 and LKB1) subtype, which was a novel subtype, had the shortest progression‐free survival (PFS) in all types of KRAS co‐mutation patients (P T type had significantly longer survival than those in KRASG>C type or KRASG>A type. Conclusion Our study revealed that concurrent genomic alterations can further stratify KRAS‐mutant lung adenocarcinoma patients into various subgroups with distinctive therapeutic responses and differential survival outcomes. The KPL is a novel and less responsive subtype among KRAS‐mutated NSCLC, and further investigation of effective treatment for this subtype is warranted., Our study revealed that concurrent genomic alterations can further stratify KRAS‐mutant lung adenocarcinoma patients into various subgroups with distinctive therapeutic responses and differential survival outcomes. The KPL is a novel and less responsive subtype among KRAS‐mutated NSCLC, and further investigation of effective treatment for this subtype is warranted.

A quasi-optical mode converter (QOMC) for a 0.42 THz, TE17,4 mode pulsed gyrotron oscillator is designed, fabricated and tested. Based on the coupled-wave theory, geometric optics theory and vector diffraction theory, a computer code is developed to design and optimize the QOMC which converts the high order TE17,4 mode into a fundamental Gaussian distribution. The numerical results show that the scalar fundamental Gaussian mode content of the wave beam at the output window is about 98.1%. The experiment result shows that a well-focused Gaussian-like beam with 95.2% scalar Gaussian mode content has been obtained by using an infrared video camera. The measured results agree well with the theoretical predictions.

A method was proposed to efficiently separate radiations from a dual-frequency operation gyrotron. The operation modes TE02 and TE04 are converted into linearly polarized Gaussian beams, and the theoretical predictions of the power conversion efficiency are 85% for TE02 mode and 90% for TE04 mode, respectively. The linearly polarized beams with different frequencies are oblique incident toward a spatial filter based on the frequency-selective surface, the beam at 0.11 THz is totally reflected, the beam at 0.22 THz is transmitted, and the transmittance is $\sim 95$ %. The transmission characteristics of the spatial filter change a little when the incident angle ranges from 0° to 30°. The preliminary experiment for the transmission coefficient of the spatial filter is carried out when the incident angle is 0°, the trend of the transmission coefficient from the theoretical predictions agrees well with the experimental results, but there is a little difference at 0.11 and 0.22 THz due to the fabrication error.

Biometric identification is an important guarantee for social security. In recent years, as the development of social and economic, the more accuracy and safety of identification are required. The person identity verification systems that use a single biometric appear inherent limitations in accuracy, user acceptance, universality. Limitations of unimodal biometric systems can be overcome by using multimodal biometric systems, which combines the conclusions made by a number of unrelated biometrics indicators. Aiming at the limitations of unimodal biometric identification, a recognition algorithm for multimodal biometric fusion based on hand vein, iris and fingerprint was proposed. To verify person identity, the hand vein images, iris images and fingerprint images were preprocessed firstly. The region of interest (ROI) of hand vein image was obtained and filtered to reduce image noises. The multiresolution analysis theory was utilized to extract the texture information of hand vein. The iris image was preprocessed through iris localization, eyelid detection, image normalization and image enhancement, and then the feature code of iris was extracted from the detail images obtained using wavelet transform. The texture feature information represented fingerprint pattern was extracted after filtering and image enhancement. The Bayesian theorem was employed to realize the fusion at the matching score level and the fusion recognition result was finally obtained. The experimental results were presented, which showed that the recognition performance of the proposed fusion method was obviously higher than that of single biometric recognition algorithm. It had verified the efficiency of the proposed method for biometrics.

The identification technology based on multi-biometric can greatly improve the applicability, reliability and antifalsification. This paper presents a multi-biometric system bases on embedded system, which includes: three capture daughter boards are applied to obtain different biometric: one each for fingerprint, iris and vein of the back of hand; FPGA (Field Programmable Gate Array) is designed as coprocessor, which uses to configure three daughter boards on request and provides data path between DSP (digital signal processor) and daughter boards; DSP is the master processor and its functions include: control the biometric information acquisition, extracts feature as required and responsible for compare the results with the local database or data server through network communication. The advantages of this system were it can acquire three different biometric in real time, extracts complexity feature flexibly in different biometrics' raw data according to different purposes and arithmetic and network interface on the core-board will be the solution of big data scale. Because this embedded system has high stability, reliability, flexibility and fit for different data scale, it can satisfy the demand of multi-biometric recognition.

As one kind of the latest forms of biometrics, the human hand vein recognition utilizes a state-of-the-art recognition algorithm based on unique veins and capillaries found on human dorsal hand, which possesses the advantages such as well anti-falsification and high noise immunity. For the hand vein recognition, the most important premise is acquiring the high quality hand vein image. According to the special effect of human hand vein on the near infrared (NIR), when a hand is scanned by an image sensor, the vein pattern appears darker than its surroundings. Depending on this characteristic, the NIR light source was utilized to illuminate the image acquisition system for hand vein. And the optimal parameters of light source were chosen and the light source with high uniformity illuminance was manufactured to acquire the more clear hand vein image. Simultaneously, for the purpose of system miniaturization and design flexibility, the embedded image acquisition system for hand vein was designed based on the technology of system on programmable chip (SOPC). FPGA and CMOS image sensor were taken as the core components in the system, and the hardware of acquisition module is realized by configuring NiosII soft-core CPU and some corresponding interface modules on a FPGA. The software was developed by using the NiosII IDE to realize the initialization control to CMOS image sensor and collection, storage and transmission for the image data gathered from CMOS. Then the collected hand vein image was simply preprocessed, which further improved the image quality. Through experiments, the results indicated that this system could obtain the hand vein image with high performance, and it supplied the embedded development platform for hand vein recognition simultaneously. It was significant to develop the hand vein recognition system with small size and high speed.

The flatness of pins is an important quality indicator for integrated circuit packaging. Almost all of the detection methods which are currently used can't be successful on efficiency and precision. In this system, the image of IC pins was captured by an properly optical systems and corresponding CCD sensor. To detect the edge of each pin, traditional algorithmic, such as Sobel operator and Roberts operator, have some disadvantages: the edge is too thick for system to accurately measure and the edge show directional character. An image segmentation and border extracting algorithm focus on the extreme of neighborhood image intensity change was adopted. The advantage of this algorithm was each pixel's neighborhood image intensity information was considered, so the algorithm is more suitable for accurately measure. After edge was extracted, how to identify the useful spots is cast as a binary classification task. The support vector machine (SVM) would be used to identify pin's spots. After proper image characteristics are obtained and a certain amount of training, SVM provides higher discrimination ratio to distinguish spots of the IC pins. To measure the flatness of pin, a particular line which can be identified easily should be put in the image as a baseline. Through calculating the distance between the pins spot and baseline, the flatness of pins is obtained accurately. In this system, the flatness of IC pins can be accurately and quickly measured, which is worthy of broad application prospect in IC packaging.

Based on the unique characteristic, the paper currency numbers can be put into record and the automatic identification equipment for paper currency numbers is supplied to currency circulation market in order to provide convenience for financial sectors to trace the fiduciary circulation socially and provide effective supervision on paper currency. Simultaneously it is favorable for identifying forged notes, blacklisting the forged notes numbers and solving the major social problems, such as armor cash carrier robbery, money laundering. For the purpose of recognizing the paper currency numbers, a recognition algorithm based on neural network is presented in the paper. Number lines in original paper currency images can be draw out through image processing, such as image de-noising, skew correction, segmentation, and image normalization. According to the different characteristics between digits and letters in serial number, two kinds of classifiers are designed. With the characteristics of associative memory, optimization-compute and rapid convergence, the Discrete Hopfield Neural Network (DHNN) is utilized to recognize the letters; with the characteristics of simple structure, quick learning and global optimum, the Radial-Basis Function Neural Network (RBFNN) is adopted to identify the digits. Then the final recognition results are obtained by combining the two kinds of recognition results in regular sequence. Through the simulation tests, it is confirmed by simulation results that the recognition algorithm of combination of two kinds of recognition methods has such advantages as high recognition rate and faster recognition simultaneously, which is worthy of broad application prospect.