Search

Linear tomographic absorption spectroscopy (LTAS) is a non-destructive diagnostic technique widely employed for gas sensing. The inverse problem of LTAS represents a classic example of an ill-posed problem. Linear iterative algorithms are commonly employed to address such problems, yielding generally poor reconstruction results due to the incapability to incorporate suitable prior conditions within the reconstruction process. Data-driven deep neural networks (DNN) have shown the potential to yield superior reconstruction results; however, they demand a substantial amount of measurement data that is challenging to acquire. To surmount this limitation, we proposed an untrained neural network (UNN) to tackle the inverse problem of LTAS. In conjunction with an early-stopping method based on running variance, UNN achieves improved reconstruction accuracy without supplementary training data. Numerical studies are conducted to explore the optimal network architecture of UNN and to assess the reliability of the early-stopping method. A comparison between UNN and superiorized ART (SUP-ART) substantiates the exceptional performance of UNN.

The monitoring of Li-ion battery temperatures is essential to ensure high efficiency and safety. In this work, two types of recurrent neural networks (RNNs), which are long short-term memory-RNN (LSTM-RNN) and gated recurrent unit-RNN (GRU-RNN), are proposed to estimate the surface temperature of 18650 Li-ion batteries during the discharging processes under different ambient temperatures. The datasets acquired from the Prognostics Center of Excellence (PCoE) of NASA are used to train, validate and test the networks. In previous work, temperature has been set as the output of the networks; however, here the temperature difference along the time axis is adopted as the output. The net heat generated results in net temperature change, which is more closely aligned with electrochemical and thermodynamic laws. Extensive simulation results show that the two RNNs can achieve accurate real-time battery temperature estimation. The maximum absolute error in temperature estimation is approximately 0.75°C and the correlation coefficient between the estimated and measured temperature curves is greater than 0.95. The influences of three crucial parameters, which are the number of hidden neurons, initial learning rate and maximum number of iterations, are also assessed in terms of training time, root mean square error and mean absolute error.

Diagnostics tools are the underpinnings for the experimental study of combustion phenomena. The inherent dynamic and three-dimensional (3-D) nature of turbulent flames has imposed strict requirements to the measurement techniques, which should provide both temporally and spatially resolved information of the target flames. Time-resolved volumetric tomography is one of such methods that meet the stringent demands of combustion diagnostics. However, this technique usually suffers from both high computational and experimental costs. This work aims to mitigate its limitations by developing a hybrid deep neural network that integrates the classical convolutional neural network with a state-of-the-art video interpolation model. Such a network can produce high frame rate 3-D flame voxels from low frame rate two dimensional (2-D) images, reducing the computational costs and at the same time relaxing the hardware requirement. Our study has shown that the temporal resolution can be enhanced by 15-fold. Thus, kilohertz (kHz)-rate flame tomography can potentially be realized with cost-effective industrial cameras. This also facilitates the study of ultra-rapid combustion phenomena, which cannot be resolved (greater than megahertz required) even with the most expensive commercial high-speed cameras. This technique has also been found to have a strong noise immunity, and acceptable results can still be obtained even when the noise level reaches 30%.

Due to the asymmetric and dynamic nature of turbulent flames encountered in practical combustors, three-dimensional measurements have been long desired to resolve the complex flame structures. Volumetric tomography has been demonstrated to be a powerful tool for this purpose. However, for practical applications, typically only a few high-speed cameras can be applied due to their high expense and the limited optical access of the combustor. Thus, the camera arrangement is of paramount importance in order to fully explore useful information of the target fields. This work proposes an optimization method for camera arrangement to minimize the correlation between the projections captured from different perspectives. Simulative studies based on computed tomography of chemiluminescence were conducted to validate the feasibility of the method. Results suggest that the optimized camera arrangement defined in the ${4}\pi $4π space can reconstruct the flame structures more faithfully than the commonly adopted distribution defined in a ${2}\pi $2π space reported in the literature for scenarios in which the optical access is constrained due to the existence of obstacles.

Nonlinear tomographic absorption spectroscopy (NTAS) is an imaging technique that inherits the advantages of both absorption spectroscopy and tomography such as species specificity, quantitative measurements, high sensitivity and capability in resolving non-uniformities of the target flow field. It has found many applications such as engine measurements where the flow field is highly turbulent. However, the solution of the NTAS problem is non-trivial as a set of nonlinear equations need to be solved. The inversion problem is typically converted into an optimization problem and solved with the simulated annealing algorithm which is criticized for its low computational efficiency. Convolutional neural networks havebeen successfully introduced to speed up the reconstructions of NTAS. However, due to the large amount of training data, CNN has a high demand for memory to store the network parameters, which also requires a large amount of time to be optimized during the learning process. In this work, we proposed a scheme to compress the output layer via dimension reduction using proper orthogonal decomposition. The simulative study performed here demonstrated that the number of network parameters and the training time can be reduced dramatically by a factor of ∼ 100 and ∼ 10 respectively. This improvement makes NTAS a more preferable technique for combustion diagnostics for which time and spatially resolved measurements are highly desired. • New compressed convolutional neural networks (CCNN) based on POD are proposed. • Comparison between CCNN and the traditional CNN is investigated, and the former outperforms the latter in terms of both computational efficiency and noise immunity. • By compressing the size of the output layer, the requirement of memory for CCNN is dramatically smaller than traditional CNN. [ABSTRACT FROM AUTHOR]

BackgroundTyrosine kinase inhibitors (TKIs) ushered in the era of targeted therapy for chronic myeloid leukemia(CML), which greatly improved the prognosis of patients with CML. Flumatinib is a second-generation tyrosine kinase inhibitor (TKI), which is currently approved for patients with Philadelphia chromosome-positive chronic myeloid leukemia (Ph+ CML) in China. And some current clinical trials revealed that flumatinib has been shown to be a more potent inhibitor of BCR::ABL than imatinib.

The plasma membrane Ca2+ ATPases (PMCAs) are responsible for the clearance of Ca2+ out of cells after intracellular Ca2+ transients. Cooperating with Na+/Ca2+ exchangers (NCXs) and Ca2+ buffering proteins, PMCAs play an essential role in maintaining the long-term cellular Ca2+ homeostasis. The plasma membrane Ca2+ ATPase was first discovered in red blood cell membrane about 50 years ago, and then other PMCA isoforms and alternatively spliced variants had been identified from different tissues and different developmental stages, revealing a surprising complexity of the PMCA family. In mammals, there are four PMCA isoforms encoded by four distinct genes. Isoform 1 and 4 are found in virtually all tissues, whereas isoform 2 and 3 are primarily expressed in excitable cells such as neurons and myocytes. Perturbation of PMCAs function has been implicated in a variety of diseases and disorders, including hearing loss, ataxia, paraplegia, and infertility. Here, we would like to review the recent progresses in the study of the PMCAs and related disorders, in particular how these pathological conditions help us to gain an in-depth insight into the function of PMCAs and their contribution in the regulation of Ca2+ signaling network.

The current standard of care for acute myeloid leukemia (AML) is largely ineffective with very high relapse rates and low survival rates, mostly due to the inability to eliminate a rare population of leukemic stem cells (LSCs) that initiate tumor growth and are resistant to standard chemotherapy. RNA-sequencing analysis on isolated LSCs confirmed C-type lectin domain family 12 member A (CLL1, also known as CLEC12A) to be highly expressed on LSCs but not on normal hematopoietic stem cells (HSCs) or other healthy organ tissues. Expression of CLL1 was consistent across different types of AML. We developed CLT030 (CLL1-ADC), an antibody-drug conjugate (ADC) based on a humanized anti-CLL1 antibody with 2 engineered cysteine residues linked covalently via a cleavable linker to a highly potent DNA-binding payload, thus resulting in a site-specific and homogenous ADC product. The ADC is designed to be stable in the bloodstream and to release its DNA-binding payload only after the ADC binds to CLL1-expressing tumor cells, is internalized, and the linker is cleaved in the lysosomal compartment. CLL1-ADC inhibits in vitro LSC colony formation and demonstrates robust in vivo efficacy in AML cell tumor models and tumor growth inhibition in the AML patient-derived xenograft model. CLL1-ADC demonstrated a reduced effect on differentiation of healthy normal human CD34+cells to various lineages as observed in an in vitro colony formation assay and in an in vivo xenotransplantation model as compared with CD33-ADC. These results demonstrate that CLL1-ADC could be an effective ADC therapeutic for the treatment of AML.

It gets more and more important for video copyright protection, it is necessary to provide efficient H.264 compression domain watermarking for video Digital rights management(DRM). A new watermarking method based on H.264 compressed domain is proposed for video DRM, in which the embedding and extracting procedure are performed using the syntactic elements of the compressed bit stream. In this way, complete decoding is unnecessary in both embedding and extracting processes. Based on the analysis of the time and space, some appropriate sub-blocks are selected for embedding watermarks, increasing watermark robustness while reducing the declination of the visual quality. In order to avoid bit-rate increasing and strengthen the security of the proposed video watermarking scheme, only a set of nonzero coefficients quantized in different parts of macroblocks is chosen for inserting watermark. The experiment results show the proposed scheme can achieve excellent robustness against some common attacks, the proposed scheme is secure and efficient for video content DRM protection.

Current methods of image protection based on chaos encryption can only provide security in Human visual system (HVS) level, however they can not provide usage control when the image is opened, printed or exported. To solve this problem, we proposed a novel image digital rights management scheme for Confidential image data security based on encryption and watermark (CIDSEW) with high-level security, usage control and traceability, in which we employed full content image encryption for confidentiality of the image to be protected, and we firstly proposed strict and detailed Usage control (UC) scheme for Confidential image data (CIData) usage password-based authentication, opening times, printing and exporting control. And when the CIData need to delivery or export to other users or domain, we proposed the secure export and misused tracing and detect approach, in which before the ciphered CIData is exported, we decrypted the CIData in a plain mode, and simultaneously we embedded user-identity-related and hardware-related information as robust watermark for traceability and responsibility confirmation. Finally, we evaluated the proposed CIDSEW by groups of variant size image data for security and efficiency, a large amount of groups of experiments manifest the proposed CIDSEW is secure, efficient, pervasive and robust for confident image data protection, usage control and misuse tracing.

Nonlinear tomography can be combined with line-of-sight measurement techniques and take the full advantage of the nonlinear relationship between the target fields to be reconstructed and the measured projections. For example, it has been integrated with absorption spectroscopy for simultaneous imaging of temperature and species concentration, and the technique is referred to as nonlinear tomographic absorption spectroscopy (NTAS) which is especially suitable to applications where the optical access is extremely limited. However, the major drawback of nonlinear tomography is its high computational costs for the inversion process, which involves the solution of a large-scale nonlinear equation system. The situation becomes more severe when thousands of tomographic frames need to be processed. This limitation can be potentially overcome by applying a deep learning algorithm, which can build efficient mapping between the projections and the target fields for rapid reconstruction. Nevertheless, only preliminary study of convolutional neural networks was reported so far, and there are also other well-established methods which have not been investigated yet. In this work, we cite NTAS as an example and demonstrate the reconstruction of temperature field using deep brief network (DBN) and recurrent neural network (RNN) for the first time. This work also aims to provide systematic comparative studies between these representative algorithms. As expected, all the deep learning algorithms are very efficient in solving NTAS problems, and a typical reconstruction time is on the order of milliseconds. In addition, the results suggest that, in NTAS problem, RNN is superior to the other methods in terms of both accuracy and noise immunity, and it is more favorable for practical applications.