Search

In this paper, we focus on the automated diagnosis of physiological curvature in the cervical spine, with an emphasis on feature point localization. Cervical spine deformity is prevalent, and the Cobb angle is widely recognized as the gold standard for diagnosing and treating it. However, manual measurement is time-consuming, labor-intensive, and heavily reliant on clinical experience. Therefore, there is an urgent need for a high-precision automatically detecting algorithm to meet the clinical requirements of orthopedic surgeons. Traditional methods are constrained by complex steps and limited data, which pose challenges. Therefore, we propose an efficient framework that formulates an automatic diagnosis of cervical spine physiological curvature based on a novel deep neural network. By leveraging the excellent properties of neural memory Ordinary Differential Equation (nmODE) in long-term memory retention and nonlinear representation capabilities, we effectively improve the network's performance in keypoint detection branching tasks. Additionally, we integrate a novel hybrid transformer based on residual structures and a multi-stage dilated dynamic convolution to alleviate false detections caused by X-ray obstruction and shadows, and the integration also captures the relationship between vertebrae and landmarks to compensate for the lack of detailed information. We constructed a dataset named CSL-947X, comprising 947 cervical spine lateral X-ray images of patients to train and evaluate our proposed model. Extensive experiments on CSL-947X demonstrate that our framework achieves higher accuracy and outperforms most state-of-the-art methods. These results highlight the effectiveness of the proposed architecture and its potential feasibility as a clinical decision-making tool for healthcare professionals. [ABSTRACT FROM AUTHOR]

The growing number of distributed renewable energy sources and dynamic constant-power loads (e.g. electric vehicle charging stations) pose new challenges for network operators. These changes result in alterations to network load profiles and load flows, leading to greater voltage volatility. One effective solution to these problems can be the use of automatic voltage regulators (AVRs), which stabilize and symmetrize voltage output, whether at distribution transformers (DTs) or elsewhere in the distribution network. The device developed by the authors consists of two bidirectional power converters and three single-phase transformers connected in series to the low-voltage grid as a stabilizer. The proposed control system provides accurate and fast regulation of the AVR output voltage (within the range of ±10% of the nominal grid voltage), with each phase being independently adjusted, regardless of the type of power load. The article includes test results demonstrating selected functionalities of the developed AVR. The physical model of the device discussed in the article is a research component of the LINTE2 laboratory of the Gdańsk University of Technology. [ABSTRACT FROM AUTHOR]

Background: Cone-beam computed tomography (CBCT) is a convenient method for adaptive radiation therapy (ART), but its application is often hindered by its image quality. We aim to develop a unified deep learning model that can consistently enhance the quality of CBCT images across various anatomical sites by generating synthetic CT (sCT) images. Methods: A dataset of paired CBCT and planning CT images from 135 cancer patients, including head and neck, chest and abdominal tumors, was collected. This dataset, with its rich anatomical diversity and scanning parameters, was carefully selected to ensure comprehensivemodel training. Due to the imperfect registration, the inherent challenge of local structural misalignment of paired datasetmay lead to suboptimal model performance. To address this limitation, we propose SynREG, a supervised learning framework. SynREG integrates a hybrid CNN-transformer architecture designed for generating high-fidelity sCT images and a registration network designed to correct local structural misalignment dynamically during training. An independent test set of 23 additional patients was used to evaluate the image quality, and the results were compared with those of several benchmark models (pix2pix, cycleGAN and SwinIR). Furthermore, the performance of an autosegmentation application was also assessed. Results: The proposed model disentangled sCT generation from anatomical correction, leading to a more rational optimization process. As a result, the model effectively suppressed noise and artifacts in multisite applications, significantly enhancing CBCT image quality. Specifically, the mean absolute error (MAE) of SynREG was reduced to 16.81 ± 8.42 HU, whereas the structural similarity index (SSIM) increased to 94.34 ± 2.85%, representing improvements over the raw CBCT data, which had the MAE of 26.74 ± 10.11 HU and the SSIM of 89.73 ± 3.46%. The enhanced image quality was particularly beneficial for organs with low contrast resolution, significantly increasing the accuracy of automatic segmentation in these regions. Notably, for the brainstem, the mean Dice similarity coefficient (DSC) increased from 0.61 to 0.89, and the MDA decreased from 3.72 mm to 0.98 mm, indicating a substantial improvement in segmentation accuracy and precision. Conclusions: SynREG can effectively alleviate the differences in residual anatomy between paired datasets and enhance the quality of CBCT images. [ABSTRACT FROM AUTHOR]

Summary: This study presents a compact hybrid dual‐frequency impedance transformer that is suitable for sources and loads having frequency‐dependent complex impedances. The basic transformer comprises two transmission lines and two shunt stubs, and this configuration is adequate for the design of a dual‐frequency impedance transformer for a few sources and loads. Compared with other studies, this work presents a distinct derivation concept to obtain the circuit parameters, thus enabling the achievement of a compact circuit. For some particular sources and loads that the basic structure is difficult to match because of line width limitations, lumped elements are added to overcome the limitation and further reduce the circuit size. The derived design formulas avoid errors caused by frequency‐dependent deviations in the component values due to parasitic effects, which constitute a critical problem in Radio Frequency (RF) circuits that employ lumped elements. The simulated results are in good agreement with the measured results, and the frequency responses obtained herein validate the proposed structure and design procedure. [ABSTRACT FROM AUTHOR]

Authorship attribution for less-resourced languages like Romanian, characterized by the scarcity of large, annotated datasets and the limited number of available NLP tools, poses unique challenges. This study focuses on a hybrid Transformer combining handcrafted linguistic features, ranging from surface indices like word frequencies to syntax, semantics, and discourse markers, with contextualized embeddings from a Romanian BERT encoder. The methodology involves extracting contextualized representations from a pre-trained Romanian BERT model and concatenating them with linguistic features, selected using the Kruskal–Wallis mean rank, to create a hybrid input vector for a classification layer. We compare this approach with a baseline ensemble of seven machine learning classifiers for authorship attribution employing majority soft voting. We conduct studies on both long texts (full texts) and short texts (paragraphs), with 19 authors and a subset of 10. Our hybrid Transformer outperforms existing methods, achieving an F1 score of 0.87 on the full dataset of the 19-author set (an 11% enhancement) and an F1 score of 0.95 on the 10-author subset (an increase of 10% over previous research studies). We conduct linguistic analysis leveraging textual complexity indices and employ McNemar and Cochran's Q statistical tests to evaluate the performance evolution across the best three models, while highlighting patterns in misclassifications. Our research contributes to diversifying methodologies for effective authorship attribution in resource-constrained linguistic environments. Furthermore, we publicly release the full dataset and the codebase associated with this study to encourage further exploration and development in this field. [ABSTRACT FROM AUTHOR]

Sentiment analysis is important for comprehending attitudes and emotions, and popular social media platform X (formally Twitter) is useful in this context. For sentiment analysis of English texts, several approaches are available but the Arabic language calls for more specialized study because of its unique qualities and subtleties. This paper presents TransNet, a Deep Attentional Hybrid Transformer model designed to classify asthma-related Arabic social media messages. TransNet combines the sequential learning capabilities of Gated Recurrent Units (GRUs) and Long Short-Term Memory Networks (LSTMs) with the resilient attention mechanisms of Transformers. The model can extract complex patterns and relationships from textual input efficiently. We use label encoding and upsampling approaches to rectify the intrinsic class imbalance in our Kaggle dataset. We also expand the dataset containing Arabic asthma-related tweets with new data, namely by substituting synonyms to add variety to the training set. To better capture the subtleties of the Arabic language, we further improve text representation by adding a pre-trained Bidirectional Encoder Representations from Transformers (BERT) tokenizer. A rigorous collection of baseline models, including Transformer+LSTM, Transformer+GRU, Transformer+CNN, and Transformer+GRU-CNN, are used to assess TransNet’s performance. TransNet demonstrates its supremacy with an F1 score of 97.86% and an excellent accuracy of 97.87% in the empirical data. We use Local Interpretable Model-agnostic Explanations (LIME), which helps to understand the mechanism of the model and also ensures that our forecasts are clear and comprehensible. According to our study, TransNet performs better than conventional models in categorizing Arabic asthma postings on social media and gives useful insights.

The unprecedented growth of distributed renewable generation is changing the distribution network from passive to active, resulting in issues like reverse power flow, voltage violations, malfunction of protection relays, etc. To ensure the reliable and flawless operation of active distribution networks, an electrical device enabling active network management is necessary, and a hybrid distribution transformer offers a promising solution. This study introduces a novel hybrid transformer topology and multi-mode control strategy to achieve coordinated voltage and reverse power regulation in active distribution networks. The proposed hybrid transformer combines conventional transformer windings with a partially rated SiC-MOSFET-based back-to-back converter, reducing additional investment costs and enhancing system reliability. A multi-mode control strategy is proposed to facilitate the concurrent reverse power control and voltage violation mitigation of the presented hybrid transformer, allowing a smooth transition between the P–Q control mode and the V–f control mode. The control mode switching can be activated manually or autonomously in response to voltage violations or reverse power overloading. The effectiveness of the proposed hybrid transformer configuration and its control mode transition mechanism are examined through comprehensive case studies conducted in the PSCAD/EMTDC environment. The proposed HT design has been confirmed to achieve a voltage regulation range of ±20% of the nominal voltage and effectively regulate bidirectional active power flow within a range of −25% to 25% of the rated power. [ABSTRACT FROM AUTHOR]

Power quality is one of the most important aspects of the power system, especially in the distribution system. This system can be affected by such disturbances as short-term voltage sags/swells, long-term voltage increases/decreases over the normative limits, harmonic distortion, interruptions, etc. Maintaining the electricity quality parameters in accordance with the applicable standards in the power distribution system will require the use of additional devices. The scientific literature on the subject indicates a very wide range of different topologies of AC voltage compensators. In this paper, different power electronics-based AC voltage compensator topologies are reviewed that are typically used with distribution energy systems. A summary of the properties and functionalities of the selected topologies has been presented. This review focuses mainly on the different topologies of the Unified Power Quality Controllers (UPQC) and Hybrid Transformers (HT). This review is divided into several sections. The first section presents the topologies used in UPQC and HT. The second section discusses selected control methods for these systems. The third section presents the basic properties of the systems related to compensating changes in the supply voltage and improvement of power quality. The fourth section presents the use of UPQC and HT systems. Finally, a summary and conclusions are presented. [ABSTRACT FROM AUTHOR]

The modern low-voltage distribution system is exposed to frequent changes in voltage amplitude due to the presence of high-power receivers with variable operating characteristics as well as distributed renewable energy sources whose generation levels depend on weather conditions. Maintaining the appropriate parameters related to power quality in the context of permissible voltage levels more and more often requires the use of additional voltage regulators. These are static systems in which the values of taps on transformers are changed, or dynamic compensators using power electronic converters. This article presents the continuation of research on one of the proposals of AC voltage compensators based on a bipolar AC/AC converter. The general properties of the presented system are reviewed and an analysis of the range of generated output voltages depending on the phase shift of the compensating voltage are presented. The next part of the article presents the formulas for calculating the duty cycle factors for the control functions of individual converters. The verification of the determined dependencies is presented on the basis of simulation tests of the proposed system. At the end of this article, the disadvantages of the proposed open-loop control system regulation and a proposal for further research are indicated. [ABSTRACT FROM AUTHOR]

Objective: Epilepsy prediction algorithms offer patients with drug-resistant epilepsy a way to reduce unintended harm from sudden seizures. The purpose of this study is to investigate the applicability of transfer learning (TL) technique and model inputs for different deep learning (DL) model structures, which may provide a reference for researchers to design algorithms. Moreover, we also attempt to provide a novel and precise Transformer-based algorithm. Methods: Two classical feature engineering methods and the proposed method which consists of various EEG rhythms are explored, then a hybrid Transformer model is designed to evaluate the advantages over pure convolutional neural networks (CNN)-based models. Finally, the performances of two model structures are analyzed utilizing patient-independent approach and two TL strategies. Results: We tested our method on the CHB-MIT scalp EEG database, the results showed that our feature engineering method gains a significant improvement in model performance and is more suitable for Transformer-based model. In addition, the performance improvement of Transformer-based model utilizing fine-tuning strategies is more robust than that of pure CNN-based model, and our model achieved an optimal sensitivity of 91.7% with false positive rate (FPR) of 0.00/h. Conclusion: Our epilepsy prediction method achieves excellent performance and demonstrates its advantage over pure CNN-based structure in TL. Moreover, we find that the information contained in the gamma ( $\gamma$ ) rhythm is helpful for epilepsy prediction. Significance: We propose a precise hybrid Transformer model for epilepsy prediction. The applicability of TL and model inputs is also explored for customizing personalized models in clinical application scenarios.

Copyright of Przegląd Elektrotechniczny is the property of Przeglad Elektrotechniczny and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Authorship attribution for less-resourced languages like Romanian, characterized by the scarcity of large, annotated datasets and the limited number of available NLP tools, poses unique challenges. This study focuses on a hybrid Transformer combining handcrafted linguistic features, ranging from surface indices like word frequencies to syntax, semantics, and discourse markers, with contextualized embeddings from a Romanian BERT encoder. The methodology involves extracting contextualized representations from a pre-trained Romanian BERT model and concatenating them with linguistic features, selected using the Kruskal–Wallis mean rank, to create a hybrid input vector for a classification layer. We compare this approach with a baseline ensemble of seven machine learning classifiers for authorship attribution employing majority soft voting. We conduct studies on both long texts (full texts) and short texts (paragraphs), with 19 authors and a subset of 10. Our hybrid Transformer outperforms existing methods, achieving an F1 score of 0.87 on the full dataset of the 19-author set (an 11% enhancement) and an F1 score of 0.95 on the 10-author subset (an increase of 10% over previous research studies). We conduct linguistic analysis leveraging textual complexity indices and employ McNemar and Cochran’s Q statistical tests to evaluate the performance evolution across the best three models, while highlighting patterns in misclassifications. Our research contributes to diversifying methodologies for effective authorship attribution in resource-constrained linguistic environments. Furthermore, we publicly release the full dataset and the codebase associated with this study to encourage further exploration and development in this field.

Multi-modal medical image fusion (MMIF) has found wide application in the field of disease diagnosis and surgical guidance. Despite the popularity of deep learning (DL)-based fusion methods, these DL algorithms cannot provide satisfactory fusion performance due to the difficulty in capturing the local information and the long-range dependencies effectively. To address these issues, this paper has presented an unsupervised MMIF method by combining a densely-connected high-resolution network (DHRNet) with a hybrid transformer. In this method, the local features are firstly extracted from the source image using the DHRNet. Then these features are input into the fine-grained attention module in the hybrid transformer to produce the global features by exploring their long-range dependencies. The local and global features are fused by the projection attention module in the hybrid transformer. Finally, based on the fused features, the fused result is reconstructed by the decoder network. The presented network is trained using an unsupervised loss function including edge preservation value, structural similarity, sum of the correlations of differences and structural tensor. Experiments on various multi-modal medical images show that, compared with several traditional and DL-based fusion methods, the presented method can generate visually better fused results and provide better quantitative metrics values. [ABSTRACT FROM AUTHOR]

This paper presents a novel hybrid transformer (HT) to regulate voltage sags/swells and to provide a unity power factor at the connection point to the medium voltage grid. The HT is a promising instrument for constructing an active distribution grid in the future. To overcome the applied design limitations concerning the converter and control system of the existing HTs, the HT structure and the control design have been proposed based on the indirect matrix converter (IMC) and Lyapunov energy function, respectively. The proposed HT topology consists of a direct power converter and a line frequency transformer. The proposed HT employs an IMC with space vector modulation that is connected to the distribution network via a series transformer. This power electronic converter increases the reliability, controllability and intelligence of the distribution substation. The Lyapunov direct control method is used to design the IMC controller that ensures tracking the input reference current and voltage. A systematic experimental and numerical investigation is performed in order to validate the effectiveness of the proposed topology. The proposed HT can compensate voltage sags/swells (more than 20% of nominal voltage). In addition, it performs properly in power factor correction and reduces the capacitor bank in the station. [ABSTRACT FROM AUTHOR]

This article proposes a new architecture for the hybrid distribution transformer (HDT) with a partially rated power converter connected in series with the existing distribution transformer. The proposed single-phase system is constructed by adding a three-legged power converter in series to the existing transformer's primary side. Here, the way the converter is connected to the transformer's two taps provides the following advantages: 1) It is possible to reduce the power converter's power rating, and 2) An additional current loop, which gives a degree of freedom, is added. For better power quality, the proposed system synthesizes a voltage vector to compensate for the voltage sags/swells and the grid voltage harmonics. Furthermore, it improves the grid's power factor to increase the power transmission efficiency. The proposed system meets the needs of smart grids in terms of resilience. The configuration, control strategies, analysis, and activation/deactivation sequence of the proposed system are discussed in this article. Real-scale simulations and lab-scale experiments were performed to validate the proposed system, and the system's effectiveness was verified through the obtained results. [ABSTRACT FROM AUTHOR]

Power quality is one of the most important aspects of the power system, especially in the distribution system. This system can be affected by such disturbances as short-term voltage sags/swells, long-term voltage increases/decreases over the normative limits, harmonic distortion, interruptions, etc. Maintaining the electricity quality parameters in accordance with the applicable standards in the power distribution system will require the use of additional devices. The scientific literature on the subject indicates a very wide range of different topologies of AC voltage compensators. In this paper, different power electronics-based AC voltage compensator topologies are reviewed that are typically used with distribution energy systems. A summary of the properties and functionalities of the selected topologies has been presented. This review focuses mainly on the different topologies of the Unified Power Quality Controllers (UPQC) and Hybrid Transformers (HT). This review is divided into several sections. The first section presents the topologies used in UPQC and HT. The second section discusses selected control methods for these systems. The third section presents the basic properties of the systems related to compensating changes in the supply voltage and improvement of power quality. The fourth section presents the use of UPQC and HT systems. Finally, a summary and conclusions are presented.

The modern low-voltage distribution system is exposed to frequent changes in voltage amplitude due to the presence of high-power receivers with variable operating characteristics as well as distributed renewable energy sources whose generation levels depend on weather conditions. Maintaining the appropriate parameters related to power quality in the context of permissible voltage levels more and more often requires the use of additional voltage regulators. These are static systems in which the values of taps on transformers are changed, or dynamic compensators using power electronic converters. This article presents the continuation of research on one of the proposals of AC voltage compensators based on a bipolar AC/AC converter. The general properties of the presented system are reviewed and an analysis of the range of generated output voltages depending on the phase shift of the compensating voltage are presented. The next part of the article presents the formulas for calculating the duty cycle factors for the control functions of individual converters. The verification of the determined dependencies is presented on the basis of simulation tests of the proposed system. At the end of this article, the disadvantages of the proposed open-loop control system regulation and a proposal for further research are indicated.

• We proposed a Hybrid Efficient Transformer (HET-RL) model for accurate diagnosis of various pulmonary diseases using multimodal data. • Dual-level pre-processing, such as denoising with Steered Filter (SF) and normalization with Min-Max normalization, improved data quality. • The HET-RL model consists of Inter-Attention Transformer (IAT) and Text Analyzer Transformer (TAT) for data analysis. • The model demonstrated superior performance, with 12% and 9% improvement in identifying pulmonary diseases compared to non-unified multimodal and image-only models. Pulmonary diseases, encompassing conditions such as chronic bronchitis, emphysema, asthma, and pulmonary fibrosis, involve intricate pathophysiological mechanisms affecting the respiratory system, necessitating precise diagnosis and tailored therapeutic approaches. Timely and accurate diagnosis of pulmonary diseases is crucial as it enables early intervention, optimal management, and prevention of complications, thereby improving patient outcomes and quality of life. The scarcity of multi-modality datasets and challenges in accurate diagnosis underscore the complexities faced by deep learning models in achieving comprehensive pulmonary diagnoses, emphasizing the need for enhanced data diversity and algorithmic robustness in addressing diagnostic issues. To overcome these challenges, we have proposed a hybrid efficient transformer based on representation learning named as HET-RL model for accurate various pulmonary disease using multi-modality. Primarily in our work, we utilized multiple data including radiograph, chief complaints and clinical parameters for enhancing the efficiency of model performance. We have performed dual-level pre-processing such as denoising and normalization for amplifying data quality using Steered Filter (SF) and Min-Max normalization, respectively. Then, we proposed HET-RL model which encompasses of Inter-Attention Transformer (IAT) and Text Analyzer Transformer (TAT) for data analyzing. In which, appropriate features are analyzed and extracted from radiograph (CT scan) by IAT and TAT with representation learning (RL) encode the text and extract significant information from both chief complaint clinical parameters. Finally, the extracted features from hybrid transformers are fused by adapting Fusion Network and then pulmonary disease are classified into multiple classes. Incorporating diverse data sources, including results of laboratory test and patient demographic characteristics, our model demonstrated superior performance compared to non-unified multimodal and an image-only model diagnosis model. It exhibited a 12% and 9% improvement in identifying pulmonary disease. Multimodal hybrid transformer-based models hold promise for streamlining patient triaging and enhancing the clinical decision-making process. [ABSTRACT FROM AUTHOR]

• Hybrid transformers can manage the voltage at point of common coupling of wind farms effectively. • The maximum tolerant connected capacity of wind farms can be increased with hybrid transformers in same grid. • The wind farms can be protected from grid fault voltages with hybrid transformers. • The proposed additional reactive power circulation suppression method ensures the full use of converters of the hybrid transformers. The rapid voltage fluctuations and even exceedances caused by power variability of wind generations significantly impede their application within the power system. This paper presents a method for replacing the electromagnetic transformer at the Point of Common Coupling of wind power delivery systems with a hybrid transformer (HT) to enhance its voltage management and grid integration capabilities. Considering the operation voltage in wind farm and scalability, a cascaded H-bridge based HT model has been established. The operation modes and a set of control strategies groups have been analyzed and proposed. The HTs operate by default in reactive power compensation mode to manage the dual-side voltage coordinately. In case of reactive power contention and circulation, this paper proposes an additional reactive power circulation suppression method. Once it exceeds the reactive voltage regulation limits due to the grid faults or changes of gird operation modes, HT adopts ratio fast regulation mode, maintaining the wind-farm side node voltage and preventing wind turbine tripping. The influence of HT leakage reactance has also been studied here. Finally, the effectiveness of the proposed method is validated in simulation, which provides a new paradigm for the operational application of HTs. [ABSTRACT FROM AUTHOR]

Self-supervised monocular depth estimation aims to infer depth information without relying on labeled data. However, the lack of labeled information poses a significant challenge to the model's representation, limiting its ability to capture the intricate details of the scene accurately. Prior information can potentially mitigate this issue, enhancing the model's understanding of scene structure and texture. Nevertheless, solely relying on a single type of prior information often falls short when dealing with complex scenes, necessitating improvements in generalization performance. To address these challenges, we introduce a novel self-supervised monocular depth estimation model that leverages multiple priors to bolster representation capabilities across spatial, context, and semantic dimensions. Specifically, we employ a hybrid transformer and a lightweight pose network to obtain long-range spatial priors in the spatial dimension. Then, the context prior attention is designed to improve generalization, particularly in complex structures or untextured areas. In addition, semantic priors are introduced by leveraging semantic boundary loss, and semantic prior attention is supplemented, further refining the semantic features extracted by the decoder. Experiments on three diverse datasets demonstrate the effectiveness of the proposed model. It integrates multiple priors to comprehensively enhance the representation ability, improving the accuracy and reliability of depth estimation. Codes are available at: https://github.com/MVME-HBUT/MPRLNet. • Exploring the complementary cues of multiple priors. • A hybrid transformer and lightweight pose network are employed to capture spatial priors. • Leveraging context priors to perceive complex structures. • Semantic priors are introduced to enhance the object representation. [ABSTRACT FROM AUTHOR]

To solve the security problem posed by fake face presentation attacks in face recognition systems, many detection methods have been proposed in recent years. Among these detection methods, the methods by analyzing a single image have achieved good real-time performance. However, their detection performances are often limited by the effectiveness of the extracted features. Compared with the real face, the fake face always shows structure and texture differences. Based on this, we propose a detection method that fuses local texture and structural depth information, where the Swin-Transformer is invoked to generate both local and constructed features. More specifically, the face image is first divided into non-overlapped small patches, which can further extract the features in micro-clues. Then, the Swin-Transformer with four stages is invoked to encode these patches, and the outputs from different stages are fused to generate constructed depth descriptors and local texture features. Finally, the local texture features and constructed depth map are combined to classify whether the input face image is captured from a real face. Extensive experiments about intra-dataset and cross-dataset evaluation indicate that our proposed approach significantly outperforms the state-of-the-art methods and show promising generalization performance. [ABSTRACT FROM AUTHOR]

Background: Cone-beam computed tomography (CBCT) is a convenient method for adaptive radiation therapy (ART), but its application is often hindered by its image quality. We aim to develop a unified deep learning model that can consistently enhance the quality of CBCT images across various anatomical sites by generating synthetic CT (sCT) images., Methods: A dataset of paired CBCT and planning CT images from 135 cancer patients, including head and neck, chest and abdominal tumors, was collected. This dataset, with its rich anatomical diversity and scanning parameters, was carefully selected to ensure comprehensive model training. Due to the imperfect registration, the inherent challenge of local structural misalignment of paired dataset may lead to suboptimal model performance. To address this limitation, we propose SynREG, a supervised learning framework. SynREG integrates a hybrid CNN-transformer architecture designed for generating high-fidelity sCT images and a registration network designed to correct local structural misalignment dynamically during training. An independent test set of 23 additional patients was used to evaluate the image quality, and the results were compared with those of several benchmark models (pix2pix, cycleGAN and SwinIR). Furthermore, the performance of an autosegmentation application was also assessed., Results: The proposed model disentangled sCT generation from anatomical correction, leading to a more rational optimization process. As a result, the model effectively suppressed noise and artifacts in multisite applications, significantly enhancing CBCT image quality. Specifically, the mean absolute error (MAE) of SynREG was reduced to 16.81 ± 8.42 HU, whereas the structural similarity index (SSIM) increased to 94.34 ± 2.85%, representing improvements over the raw CBCT data, which had the MAE of 26.74 ± 10.11 HU and the SSIM of 89.73 ± 3.46%. The enhanced image quality was particularly beneficial for organs with low contrast resolution, significantly increasing the accuracy of automatic segmentation in these regions. Notably, for the brainstem, the mean Dice similarity coefficient (DSC) increased from 0.61 to 0.89, and the MDA decreased from 3.72 mm to 0.98 mm, indicating a substantial improvement in segmentation accuracy and precision., Conclusions: SynREG can effectively alleviate the differences in residual anatomy between paired datasets and enhance the quality of CBCT images., Competing Interests: The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2024 Hu, Cheng, Wei and Liang.)

This article proposes a dual-band complementary metal–oxide–semiconductor (CMOS) radio frequency (RF) front end employing a hybrid transformer-based electrical-balance duplexer (EBD) and $N$ -path filter embedded low-noise amplifier (LNA) for both in-band full duplexing (IBFD) and frequency-division duplexing (FDD) radios. To securely obtain sufficient self-interference cancellation (SIC) or transmitter (TX) leakage rejection in IBFD and FDD modes of the EBD at both low-band (LB) (0.7–1 GHz) and mid-band (MB) (1.8–2.2 GHz) frequency ranges, a band-switchable hybrid transformer is designed with a finely adjustable balance network to control the antenna impedance variations. In addition, the four-phase $N$ -path LNA is utilized to further offer TX leakage rejection in the FDD operation. This RF front-end is fabricated using a 65-nm CMOS technology and is mainly characterized by long-term evolution frequency bands. The demonstrated design achieves SIC and TX leakage rejection greater than 55 dB for IBFD and FDD modes of all measured frequency bands, which has a maximum power handling capability of +25 dBm. Furthermore, the design attains voltage gains greater than 8.7 dB for LB and 3.7 dB for MB, noise figures of 10.4 dB for LB and 12.1 dB for MB, out-of-band (OB) P1dBs greater than +21.4 dBm for LB and +21.5 dBm for MB, and OB IIP3s greater than +39.7 dBm for LB and +43.1 dBm for MB. It draws a bias current of 23.8 mA from the LNA and output buffer with a nominal supply of 1.2 V and has an active area of approximately 2 mm2. [ABSTRACT FROM AUTHOR]

A 28-GHz in-band full-duplex (IBFD) power amplifier (PA) is presented which integrates a PA with an impedance matched hybrid transformer in a 28-nm bulk CMOS process. The hybrid transformer has a 1:2 turn ratio to match from the optimum load impedance of $25~\Omega $ for a PA to the antenna and balance ports of $100~\Omega $. Since this allows the additional matching circuits of a PA and an antenna to be obviated, the full-duplex loss is reduced. It shows 28.4-dB isolation between the PA and LNA ports at 28 GHz. For comparisons, a 50- $\Omega $ matched PA and an IBFD PA are designed. The IBFD PA has core sizes of $0.24\times 0.43\,\,{\text {m}}{\text {m}}^{2}$. Since the impedance matched hybrid transformer has the similar size to the output matching network of the 50- $\Omega $ matched PA, a full duplexer is integrated into the PA without additional chip area. The IBFD PA has 10.6-dB gain and a drain efficiency of 8.9% at 12.3-dBm P1dB. [ABSTRACT FROM AUTHOR]

Copyright of Electric Power Automation Equipment / Dianli Zidonghua Shebei is the property of Electric Power Automation Equipment Press and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

This article presents a study of an AC voltage stabilizer based on a three-phase hybrid transformer combined with a matrix converter. The proposed solution is used to control AC voltage amplitude and phase shift. By adjustment of these voltage parameters we can reduce the effects of overvoltage, voltage dips or lamp flicker. Such negative phenomena are very significant, particularly from the perspective of the final consumer and sensitive loads connected to the power network. Often the voltage in the power system can be adjusted using a mechanical or thyristor controlled regulator, which in a stepwise manner switches the taps of the electromagnetic transformer. The method for obtaining continuous control of the voltage magnitude and phase shift with the use of a conventional transformer with two output windings and a matrix converter is presented in this paper. The operating principles, mathematical model and properties of the proposed voltage stabilizers are discussed in this paper. The main part of the article will be devoted to the mathematical model which is based on an averaged equation. Computer simulation results are presented and compared with the results of a mathematical study.

According to the incomplete technical features of the traditional low-frequency transformer (LFT) and the existing power electronic transformer (PET), especially in terms of the function, the voltage level, and the cost, the core flux modulation of the hybrid transformer was achieved through the establishment of auxiliary windings combined with connection of the power electronic converter (PEC). For the hybrid transformer, a three-port converter topology is proposed and the strategy applicable for the dc bias compensation by connecting the converter is presented. By utilizing the dc modulation magnetic flux, the dc bias problem could be directionally compensated. The exciting current could be improved significantly and maintained at a steady state. The method proposed does not change the connection of the transformer external circuit and does not increase the peripheral device of the transformer, which provides a new way to solve the problems of magnetic state fluctuation and excitation current distortion. [ABSTRACT FROM AUTHOR]

Graphical abstract Highlights • An idea of model predictive control realization for an AC voltage stabilizer based on hybrid transformer with a matrix converter are presented. • The compensation of voltage sags, swells and other distortion in the AC power grid are possible with the help of the proposed control strategy. • Exemplary results of simulation tests have been presented and discussed. Abstract The predictive control model is more and more frequently used in power electronics systems. It can also be used in systems to improve power quality. This article presents the concepts of a model predictive control hybrid transformer (HT) with matrix converter (MC). The proposed HT consists of a power transformer connected to a power electronic converter which controls the output voltage on the secondary side. The very simple realization and possibility to optimize various predefined criteria are advantages of the proposed control method. The compensation of voltage sags, swells and harmonics in the power grid have been observed in simulation studies. The obtained results indicate a high potential for the proposed method and significantly better performance compared to the classical control methods based on space vector modulation. A valuable contribution of the article is the use of a system with a model predictive method to compensate for stochastic distortions of the grid voltage. [ABSTRACT FROM AUTHOR]

A low-power impedance detection method using a hybrid transformer and its analysis are presented. A balun is reused for the hybrid transformer to achieve area-efficiency. The complex impedance detection enables faster optimization on a tunable matching network and its calibration technique increases the detection accuracy. The proposed impedance detection method fabricated in a 40-nm CMOS process consumes only 0.83 mW and demonstrates the accuracy of less than 18° and 0.1 on the Γ detection up to $\vert \Gamma _{{\rm{Ant}}}\vert $ of 0.5 in ISM2.4 GHz band. After single-step matching network tuning, PA power and RX noise figure are improved by up to 1.2 dB and 1.3 dB, respectively. [ABSTRACT FROM PUBLISHER]

This paper proposes an adaptive RF front-end (RFFE) architecture that uses an electrical-balance duplexer (EBD) and tuned surface acoustic wave (SAW) resonators to enable adaptive dual-frequency isolation suited for LTE's frequency-division duplex (FDD) mode of operation. Alternatively, in the in-band full-duplex (IBFD) mode, the EBD cancels the in-band TX self-interference directly at RF for increased channel capacity. Furthermore, the EBD's balance network is optimized to reduce the TX insertion loss (IL) below the nominal 3-dB limit in the FDD mode. A 0.18-μm SOI CMOS prototype implements the EBD, low-noise amplifier, and SAW-tuning capacitor banks. The chip is mounted to a module substrate together with the SAWs and matching components. The RFFE achieves >50-dB dual-frequency isolation and 2.6-3.4-dB TX IL in FDD mode, has +58/+42-dBm TX/RX-path IIP3 for an increased IBFD-link budget and handles up to +27.5-dBm TX power in both modes. [ABSTRACT FROM AUTHOR]

An electrical-balance duplexer achieving the state-of-the-art linearity and insertion loss (IL) performance is presented, enabled by a partially depleted RF silicon-on-insulator CMOS technology. A single-ended configuration avoids the common-mode isolation problem suffered by topologies with a differential low-noise amplifier. Highly linear switched capacitors allow for impedance balancing to antennas with <1.5:1 voltage standing wave ratio from 1.9 to 2.2 GHz. +70-dBm input-referred third-order intercept point is achieved under high transmitter (TX) power (+30.5 dBm max.). TX IL is <3.7 dB, and receiver IL is <3.9 dB. [ABSTRACT FROM PUBLISHER]

A hybrid transformer-based CMOS tunable duplexer with a single-ended blocker-tolerant low-noise amplifier (LNA) is proposed for a highly integrated reconfigurable RF front-end architecture. The proposed LNA adopts $Q$ -enhanced $LC$ notch filter at the source of the cascode device and $LC$ bandpass filter at the load. It improves the blocker tolerance and linearity of the receiver by rejecting unwanted out-of-band blockers and transmitter (TX) leakage signals. The duplexer with the notch-filtered LNA was fabricated in a 65-nm CMOS process. It has a voltage gain of 24.2 dB from the antenna to the LNA output with 28-dB notch filtering, cascaded noise figure of 6.4 dB, TX insertion loss of 3.8 dB, and IIP3FDof 52.4 dBm. It can also cover a mid-band range from 1.6 to 2.2 GHz in cellular applications. It draws an average current of 8 mA from a supply voltage of 1.2 V and has an active area of 1.19 mm2. [ABSTRACT FROM AUTHOR]

This paper presents a nonisolated, high boost ratio dc-dc converter with the application for photovoltaic (PV) modules. The proposed converter utilizes a hybrid transformer to incorporate the resonant operation mode into a traditional high boost ratio active-clamp coupled-inductor pulse-width-modulation dc-dc converter, achieving zero-voltage-switching (ZVS) turn-on of active switches and zero-current-switching turn-off of diodes. As a result of the inductive and capacitive energy being transferred simultaneously within the whole switching period, power device utilization (PDU) is improved and magnetic utilization (MU) is optimized. The improved PDU allows reduction of the silicon area required to realize the power devices of the converter. The optimized MU reduces the dc-bias of magnetizing current in the magnetic core, leading to smaller sized magnetics. Since the magnetizing current has low dc-bias, the ripple magnetizing current can be utilized to assist ZVS of main switch, while maintaining low root-mean-square (RMS) conduction loss. The voltage stresses on the active switches and diodes are maintained at a low level and are independent of the wide changing PV voltages as a result of the resonant capacitor in series in the energy transfer loop. The experimental results based on 250 W prototype circuit show 97.7% peak efficiency and system CEC efficiencies greater than 96.7% over 20 to 45 V input voltages. Due to the high efficiency over wide power range, the ability to operate with a wide variable input voltage and compact size, the proposed converter is an attractive design for PV module applications. [ABSTRACT FROM AUTHOR]

An RF duplexer has been fabricated on a CMOS IC for use in 3G/4G cellular transceivers. The passive circuit sustains large voltage swings in the transmit path, and isolates the receive path from the transmitter by more than 45 dB across a bandwidth of 200 MHz in 3G/4G bands I, II, III, IV, and IX. A low noise amplifier embedded into the duplexer demonstrates a cascade noise figure of 5 dB with more than 27 dB of gain. The duplexer inserts 2.5 dB of loss between power amplifier and antenna. [ABSTRACT FROM PUBLISHER]

This paper presents a nonisolated, high boost ratio hybrid transformer dc–dc converter with applications for low-voltage renewable energy sources. The proposed converter utilizes a hybrid transformer to transfer the inductive and capacitive energy simultaneously, achieving a high boost ratio with a smaller sized magnetic component. As a result of incorporating the resonant operation mode into the traditional high boost ratio pulsewidth modulation converter, the turn-off loss of the switch is reduced, increasing the efficiency of the converter under all load conditions. The input current ripple and conduction losses are also reduced because of the hybrid linear-sinusoidal input current waveforms. The voltage stresses on the active switch and diodes are maintained at a low level and are independent of the changing input voltage over a wide range as a result of the resonant capacitor transferring energy to the output of the converter. The effectiveness of the proposed converter was experimentally verified using a 220-W prototype circuit. Utilizing an input voltage ranging from 20 to 45 V and a load range of 30–220 W, the experimental results show system of efficiencies greater than 96% with a peak efficiency of 97.4% at 35-V input, 160-W output. Due to the high system efficiency and the ability to operate with a wide variable input voltage, the proposed converter is an attractive design for alternative low dc voltage energy sources, such as solar photovoltaic modules and fuel cells. [ABSTRACT FROM AUTHOR]

A tunable transformer duplexer having two notches has been proposed and demonstrated for a broadband operation. The suggested duplexer using GaAs pseudomorphic high electron mobility transistor (pHEMT) technology shows isolations of over 55 dB at the transmitter (Tx) band and over 45 dB at the receiver (Rx) band across 145 MHz bandwidth. The insertion losses are −3.35 and −3.52 dB for the Tx band and the Rx band, respectively, for 500 MHz bandwidth. At all ports, the return losses are under −15 dB. The duplexer is tunable simply by tuning the four components. Although this broadband duplexer has a very compact size with tunable capability, its performance is comparable with the front end module integrated duplexer. [ABSTRACT FROM AUTHOR]