Pedicle screw placement (PSP) is a challenging procedure in spine surgery, due to poor visualization of the vertebrae and blood vessels. This has led to the development of navigation systems that improve safety and clinical outcomes. Navigation further enabled minimal invasive approaches for PSP. Current navigation systems require a preoperative computed tomography (CT) scan, followed by manual planning by the surgeon to decide screw trajectories. Intraoperative fluoroscopy imaging may be employed to register the preoperative plan with the patient's anatomy frame or even design screw trajectory planning in real-time for each pedicle individually. This is a rather lengthy procedure that involves harmful radiation to which the surgeon is exposed repeatedly. This article explores a novel approach for PSP guidance that replaces manual planning while solely relying on 3D ultrasound (US) reconstruction, hence reducing the need for radiation. Based on an intraoperative US reconstruction, a set of anatomical landmarks of the spine are identified. Subsequently, a coordinate frame is computed per vertebra. The screw path is then generated automatically using the learned relations between the landmarks and optimal screw paths. A five-fold cross-validation is conducted on the posterior surface of CT spine data involving 90 PSP paths. The algorithm is then tested on another 50 PSP paths of the 3D-printed spine phantoms corresponding to human spine CT data. Results show that for 4 mm and 6 mm diameter screws, 98% and 84% of the computed trajectories are within the required surgical precision, respectively. Using the US-based path planning led to-0.34 ± 3.66 • and-0.45 ± 4.32 • orientation errors in the sagittal and axial planes, respectively.

Spinal diseases such as spinal degeneration and scoliosis might require pedicle screw placement (PSP) as a crucial step during surgical interventions, depending on the severity. This procedure requires the drilling of a hole for placing the screw. Thanks to imaging modalities such as computed tomography and intraoperative fluoroscopy, moving from an open approach to minimally invasive surgery (MIS) has been possible and has reduced patient complications after surgery. Still, it suffers from a lack of visual feedback for the surgeon, whereas robotic-assisted spine surgery combined with such imaging modalities can improve surgical outcomes for PSP. Yet, in such MIS procedures, physical motion, such as breathing motion, can induce shifts and deformations in the spine, leading to operation errors of approximately 2-3 mm [1]. In order to correctly identify the entry point, breathing motion needs to be compensated for. External sensors, such as an optical tracking system or range imaging, can be used to measure the motion of the skin or neighboring vertebrae, close to the entry point [2], [3]. However, Saghbiny et al. showed that the amplitude of breathing motion changes over vertebrae and has a variation of 68% from the lumbar to thoracic vertebrae [4]. Therefore, this work develops an approach for estimating the motion of each entry point, enhancing the accuracy. The proposed method uses a long short-term memory network (LSTM) on top of the inner control loop to estimate breathing motion parameters for each pedicle drilling individually, and its output is utilized to update the motion model for motion compensation during robot-assisted drilling for MIS-PSP.

Ultrafast plane-wave ultrasound imaging is a versatile tool which has become increasingly relevant for blood flow imaging using speckle tracking, but suffers from a low signal-to-noise ratio (SNR). Cascaded dual-polarity waves (CDW) imaging can improve the SNR by transmitting pulse trains, which are subsequently decoded to recover the imaging resolution. However, the current decoding method (in the time domain) requires a set of two acquisitions, which introduces motion artifacts that result in incorrect speckle tracking at high flow velocities. Here, we propose a frequency-domain decoding approach that decodes acquisitions independently. Experiments using a disk phantom show that frequency-domain decoding of a four-pulse train achieves an SNR gain of up to 4.2 dB, versus 5.9 dB for conventional decoding. The benefit of frequency-domain decoding for flow quantification is assessed through experiments performed with a rotating disk phantom and a parabolic flow, and through matching linear simulations. Both CDW methods improve the tracking accuracy compared to single plane-wave imaging. Timedomain decoding outperforms frequency-domain decoding in low SNR conditions and low velocities (≤0.25 m/s), as a result of the higher SNR gain. In contrast, frequency-domain decoding outperforms time-domain decoding for high peak velocities in imaging of the rotating disk (1 m/s) and of the parabolic flow (2 m/s), when significant scatterer motion between acquisitions causes imperfect time-domain decoding. Its ability to decode individual acquisitions makes the proposed frequency-domain decoding of CDW a promising approach to improve the SNR and thereby the accuracy of flow quantification at high velocities.

As launch and manufacturing costs have become affordable, the industrial and academic interest in Low-Earth Orbit (LEO) satellites has increased in recent years. With this interest, the concept of Low-Earth Orbit-based Positioning, Navigation and Timing (LEO-PNT) has also gained popularity as a complementary and/or standalone system in addition to the already existing Global Navigation Satellite Systems (GNSS). This article examines the LEO constellation optimization from the perspective of a LEO-PNT design, identifies and discusses the state-of-art of the LEO satellite constellation optimization approaches, introduces relevant performance and feasibility related metrics and parameters, and addresses key concepts and underlying trade-offs that must be considered for any LEO-PNT system design. In addition, a case study for a LEO-PNT constellation optimization is presented, where we showcase the discussed trade-offs. We present optimization results obtained with the adaptive weighting algorithm "ADaW" applied to the Pareto-optimization algorithm "NSGA-III". A detailed performance analysis is done for six relevant scenarios with varying receiver location properties, namely by considering indoor/outdoor, rural/urban and line of sight/non-line of sight cases.

Addressing climate change requires a transition to renewable energy sources. An integral part of this transition is the conversion of carbon dioxide through methanation, facilitated by the Power-to-X reactor technology. However, the implementation of this process can present significant operational hurdles, particularly in terms of control. The latter is critical to accommodate intermittent energy inputs. These challenges can be best addressed by model-based strategies. Traditional mechanistic models, while insightful, are constrained by computational limitations, highlighting the need for surrogate models. Classical machine learning techniques are indeed promising and have gained prominence, but still face some inherent obstacles. This study explores the capabilities of several (scientific) machine learning methods to develop a digital twin of a catalytic CO2 methanation reactor. Specifically, we investigate and compare Graph Neural Networks (GNNs), Operator Inference (OpInf), and Sparse Identification of Nonlinear Dynamics (SINDy) as potential surrogate modeling techniques, with the goal of combining data-driven approaches with underlying physical insights. These techniques will be compared with regard to their effectiveness in refining the dynamic operation of a CO2 methanation reactor using data-driven models derived from mechanistic frameworks. The goal is to provide interpretable, insightful mechanistic solutions that improve the efficiency and control of renewable energy conversion processes. 

Machine learning for optimization of the physical layer is currently a popular research topic. To aid research in this field, we introduce our Python library MOKka. We summarize the currently available signal processing modules in the library and explain our design rationale. In order to showcase the utility of this library, we have implemented a demo on joint geometric and probabilistic constellation shaping with a switchable channel model and interactive plotting and controls.

This paper presents a massive multiple-inputmultiple-output (mMIMO) system that utilizes spectrally efficient frequency division multiplexing (SEFDM) and integrates deep learning (DL) for improved channel estimation in the mMIMO-SEFDM. In contrast to the literature, this study utilizes the estimated channel feedback to dynamically adapt the SEFDM signal characteristics at the transmitter, which improves the system's agility. This enables the selection of the SEFDM signal's compression value and modulation order depending on the channel conditions, thus enhancing the symbol error rate (SER). The simulation findings demonstrate that using DLbased channel estimation and adaptive parameter selection for SEFDM leads to a better symbol error rate and improved spectral efficiency compared to orthogonal frequency division multiplexing (OFDM).

This paper presents an innovative framework for enhancing vehicular multimedia communication through adaptive computing offloading and dynamic radio access control, targeting the unique challenges of vehicular environments such as high mobility and variable network conditions. Our approach optimizes multimedia content delivery by intelligently managing computing resources and network bandwidth, focusing on maintaining high quality of service (QoS). The system uses advanced algorithms to make real-time decisions for offloading computational tasks and prioritizing multimedia traffic, ensuring efficient resource utilization and improved communication reliability. We validate our framework with simulations that demonstrate significant performance benefits over existing methods, offering potential enhancements in traffic safety and passenger multimedia experience.

This comprehensive review delves into the contemporary landscape of gas flotation in the oil and gas sector, shedding light on the evolving demands for increased separation efficiency within the industry. The oil and gas sector is now placing a premium on enhanced separation efficiency, surpassing the conventional methods currently deployed. A critical exploration of the underlying principles governing the design of high-efficiency gas flotation systems is also a focal point of this review. Gas flotation, a pivotal technology within the oil and gas industry, has witnessed a significant evolution in response to the escalating need for improved separation efficiency. This heightened demand for efficiency stems from the industry’s commitment to maximizing resource recovery, minimizing environmental impact, and optimizing operational costs. As the industry continues to evolve, there is a growing recognition of the imperative to surpass existing standards and push the boundaries of gas flotation technology. The review delves into the core principles that underpin the design of high-efficiency gas flotation systems, emphasizing the need for innovative approaches to meet the industry’s escalating demands. By scrutinizing the major design principles, researchers aim to uncover novel strategies that can revolutionize gas flotation processes, paving the way for higher performance and enhanced operational outcomes. Exploring the intricate balance between operational efficiency, environmental sustainability, and cost-effectiveness, this review serves as a roadmap for industry stakeholders seeking to navigate the shifting landscape of gas flotation technology. By synthesizing the latest advancements and industry trends, the review aims to catalyze discussions and drive innovation in the realm of high-efficiency gas flotation within the oil and gas sector.This comprehensive review further illustrates how the manipulation of gas bubble size and the intricate fluid dynamics within separation vessels serve as pivotal mechanisms for augmenting oil removal efficiency within the industry. By delving into these fundamental aspects, the review elucidates how precise control over gas bubble size and a deep understanding of fluid dynamics play a central role in enhancing the effectiveness of oil removal processes. The review goes beyond theoretical discussions to showcase real-world applications of this cutting-edge gas flotation technology within the oil and gas sector. By highlighting examples of this new generation of gas flotation technology, the review offers tangible insights into how industry players are harnessing innovative approaches to optimize oil removal efficiency. These case studies and examples serve as practical illustrations of the transformative potential of leveraging advanced gas flotation techniques to achieve higher separation efficiency and operational excellence. Through a blend of theoretical underpinnings and practical demonstrations, this review bridges the gap between theory and application, providing a holistic view of the transformative impact of gas bubble size control and fluid dynamics optimization on oil removal efficiency. By showcasing how these fundamental mechanisms translate into tangible operational improvements, the review empowers industry professionals to embrace and implement the latest advancements in gas flotation technology for enhanced oil removal performance. In essence, this review not only underscores the critical roles of gas bubble size control and fluid dynamics in boosting oil removal efficiency but also offers concrete examples of how these principles are being leveraged in practice. By showcasing the innovative strides made in the realm of gas flotation technology, the review paves the way for industry stakeholders to embrace a new era of efficiency, sustainability, and performance excellence in oil and gas separation processes.

This research investigates the modernization of inspection and monitoring practices for pipeline, casing, and tubing. The study focuses on the development and implementation of advanced inspection techniques to detect defects and imperfections such as cracks, dents, and diameter reduction, which can have devastating consequences for human safety, the environment, and the soil. The research aims to establish a comprehensive framework for the modernization of inspection and monitoring practices, encompassing the latest technologies and data analytics to ensure the integrity and reliability of pipeline, casing, and tubing infrastructure. The outcome of this research will provide a benchmark for the minimum inspection requirements, setting a new standard for the industry to ensure safe and sustainable operations.

LiDAR sensor data is essential for autonomous vehicle navigation, traffic flow monitoring, obstacle detection, and passenger safety. However, the reliability of LiDAR data can be compromised by anomalies caused by sensor malfunctions, environmental conditions, or unexpected road events. To address this, detecting anomalies in spatial-temporal (ST) LiDAR data is critical for ensuring safety. This paper proposes a novel low-complexity unsupervised framework named CNN-BiLSTM VAE for anomaly detection (AD) in non-image LiDAR data. The framework combines variational auto-encoder (VAE) reconstruction, CNN for spatial learning, and bidirectional LSTM for time-series learning in a mirror-to-mirror (M2M) architecture. Experimental results show that this method effectively detects anomalies in multidimensional ST LiDAR data, thereby maintaining robustness under various environmental conditions.

In this article, we employ active simultaneously transmitting and reflecting reconfigurable intelligent surfaces (ASRIS) to enhance the quality of 6G cellular network services. The network integrates commensal symbiotic radio (CSR) subsystems to facilitate communication between passive Internet of Things (IoT) users and active users, referred to as symbiotic backscatter devices (SBDs) and symbiotic user equipments (SUEs), respectively. Since the SBDs are passive, transmitting information to the SUEs poses significant challenges. To overcome this challenge, we harness the capabilities of massive multiple input multiple output (MIMO) antennas within the base station (BS) to relay the information transmitted by SBDs with greater power. This scheme uses the non-orthogonal multiple access (NOMA) technique for multiple access among all users, and potential interferences are eliminated using successive interference cancellation (SIC). The primary objective is to maximize the sum throughput between SBDs and SUEs. To achieve this, we formulate an optimization problem involving variables such as active beamforming coefficients at the BS and ASRIS, phase adjustments of ASRIS, and scheduling parameters between CSR and cellular networks. To solve this optimization problem, we used three deep reinforcement learning (DRL) methods: proximal policy optimization (PPO), twin delayed deep deterministic policy gradient (TD3), and asynchronous advantage actor critic (A3C). These methods were simulated, and the results demonstrate that A3C, TD3, and PPO have the best convergence speeds and achieve the highest increases in network throughput, respectively. Finally, the proposed scheme was evaluated using passive simultaneously transmitting and reflecting RIS (STAR-RIS), which demonstrated poorer performance compared to ASRIS.

Reconfigurable intelligent surfaces (RISs) are a transformative technology for improving wireless communication and localization accuracy. This paper focuses on optimizing RIS placement in the near-field (NF) scenario to minimize the position error bound (PEB) in single-input single-output (SISO) systems. Closed-form expressions for the fisher information matrix (FIM) and the PEB have been derived to evaluate localization performance. The analysis reveals that the RIS placement significantly affects localization accuracy. Two optimization methods are explored to identify the optimal RIS placement: an exhaustive search and a heuristic gray wolf optimizer (GWO). The exhaustive search method guarantees finding the optimal placement by evaluating all possible configurations discretely, but it is computationally intensive. The GWO method offers a more computationally efficient, continuous approach by imitating the behavior of gray wolves. Simulation results show that both methods effectively enhance localization accuracy, with the GWO method achieving superior performance in terms of computational efficiency and PEB minimization. The study highlights that strategic RIS placement can lead to substantial improvements in localization precision, making RIS a vital component for future wireless communication networks. This work provides a comprehensive framework for optimizing RIS deployment in NF communication scenarios, offering practical insights for enhancing localization accuracy in practical applications.

The low sensitivity of conventional single photon emission computed tomography (SPECT) presents challenges for ultra-low activity imaging, such as targeted alpha therapy (TAT) imaging. Our previous work showed that collimatorless imaging can achieve remarkable sensitivity but suffers from deteriorated resolution and inaccurate reconstructions, even with advanced reconstruction algorithms. To address these limitations, in this study, we design the dual-layer collimatorless SPECT systems specifically for fast or ultra-low activity imaging. These dual-layer systems combine collimatorless imaging with collimated acquisition using slit or slat collimators: the inner layer maximizes sensitivity with collimatorless imaging, while the outer layer provides additional spatial information through collimated imaging. Unlike general-purpose SPECT systems, these are tailored to significantly improve sensitivity while maintaining satisfactory spatial resolution and accurate reconstruction for fast or ultra-low activity imaging. In this study, our primary focus is on small animal imaging, particularly preclinical TAT imaging using 225 Ac as an exemplar for ultra-low activity imaging. We evaluate the performance of these dual-layer systems through extensive Monte Carlo simulations and reconstructions with various phantoms. The results show that relying solely on collimatorless imaging from the inner layer or collimated imaging from the outer layer leads to inaccuracies and suboptimal reconstructions in ultra-low activity imaging. The combined imaging from both layers enhances sensitivity, spatial resolution, and overall image quality. The proposed dual-layer collimatorless SPECT systems achieve a sensitivity higher than 29% for 218 keV gamma rays and an image resolution of about 5 mm, making them promising and suitable for ultra-low activity imaging.

Answer assessment is a critical aspect of various fields including education, artificial intelligence, and natural language processing. This paper aims to provide a comprehensive overview of answer assessment, exploring its methodologies, applications, challenges, and future directions. By examining the different approaches to evaluating responses, both human and automated, this paper highlights the significance of accurate and reliable answer assessment in enhancing learning outcomes, improving automated systems, and advancing research in related areas.

The exponential growth of textual data across various domains necessitates the development of efficient and accurate summarization techniques to facilitate quick comprehension and information retrieval. The novel automated system for summarizing multiple document abstracts and titles using advanced neural architectures addresses this need by leveraging large language models to generate concise and coherent summaries. The methodology involved comprehensive data collection, preprocessing, model selection, and summarization processes, evaluated through a combination of quantitative and qualitative metrics. Results demonstrated high efficacy in handling shorter documents and strong performance in technical domains such as healthcare and science, although challenges in coherence and readability were noted. Domain-specific performance highlighted the necessity for tailored adaptations, and the study contributed valuable insights into hybrid summarization techniques combining extractive and abstractive methods. Future research directions include the development of advanced attention mechanisms, domain-specific fine-tuning, and reinforcement learning techniques to optimize summarization quality, alongside addressing ethical considerations to ensure responsible deployment.

A document by Hadi Salloum. Click on the document to view its contents.

Distributed Generative Adversarial Networks (D-GANs) are a recent addition to generative model literature, but they come with significant limitations and caveats. This paper categorizes and analyzes the existing D-GAN models and introduces a novel four-part categorization framework based on their mathematical properties. We provided generalized mathematical formulations for each D-GAN category, addressing convergence issues and potential improvements, which are essential for developing more robust and reliable D-GANs. Additionally, we conducted detailed mathematical analyses of existing D-GANs to identify their shortcomings in terms of expanse, convergence and output quality. This rigorous analysis is pivotal for advancing the field, as it highlights critical areas needing improvement and sets the stage for future innovations that will enhance the performance and reliability of distributed generative models. This work establishes a foundation for future research, and standardizes the evaluation of D-GAN models. Additionally, it guides the research community towards addressing unresolved issues and exploring new avenues, thereby enhancing the overall effectiveness and applicability of the distributed generative models in AI/ML.