Search

In recent years, the growing pervasiveness of wearable technology has created new opportunities for medical and emergency rescue operations to protect users' health and safety, such as cost-effective medical solutions, more convenient healthcare and quick hospital treatments, which make it easier for the Internet of Medical Things (IoMT) to evolve. The study first presents an overview of the IoMT before introducing the IoMT architecture. Later, it portrays an overview of the core technologies of the IoMT, including cloud computing, big data and artificial intelligence, and it elucidates their utilization within the healthcare system. Further, several emerging challenges, such as cost-effectiveness, security, privacy, accuracy and power consumption, are discussed, and potential solutions for these challenges are also suggested.

Recent years have witnessed the increasing popularity and dramatic progress of contactless sensing technologies, which are able to conduct remote signal acquisition without body contact. Both the physical signs and the physiological parameters can be acquired with contactless sensing. This article introduces popular contactless sensing technologies, explores their application scenarios, and delves into the underlying theoretical principles. It comprehensively reviews the open data sets released in this field, encompassing collection scenarios, sample counts, data formats, and volunteer information. The performance baseline, typical work, and accessible links are also furnished. In addition, it includes discussions on the primary challenges and potential solutions in the context of contactless sensing with open data sets. Finally, suggestions for establishing a high-quality data set are also given to the community.

Most of the neural network architectures are based on human experience, which requires a long and tedious trial-and-error process. Neural architecture search (NAS) attempts to detect effective architectures without human intervention. Evolutionary algorithms (EAs) for NAS can find better solutions than human-designed architectures by exploring a large search space for possible architectures. Using multiobjective EAs for NAS, optimal neural architectures that meet various performance criteria can be explored and discovered efficiently. Furthermore, hardware-accelerated NAS methods can improve the efficiency of the NAS. While existing reviews have mainly focused on different strategies to complete NAS, a few studies have explored the use of EAs for NAS. In this paper, we summarize and explore the use of EAs for NAS, as well as large-scale multiobjective optimization strategies and hardware-accelerated NAS methods. NAS performs well in healthcare applications, such as medical image analysis, classification of disease diagnosis, and health monitoring. EAs for NAS can automate the search process and optimize multiple objectives simultaneously in a given healthcare task. Deep neural network has been successfully used in healthcare, but it lacks interpretability. Medical data is highly sensitive, and privacy leaks are frequently reported in the healthcare industry. To solve these problems, in healthcare, we propose an interpretable neuroevolution framework based on federated learning to address search efficiency and privacy protection. Moreover, we also point out future research directions for evolutionary NAS. Overall, for researchers who want to use EAs to optimize NNs in healthcare, we analyze the advantages and disadvantages of doing so to provide detailed guidance, and propose an interpretable privacy-preserving framework for healthcare applications.

Multiple brain imaging modalities can provide complementary pathologic information for clinical diagnosis. However, it is huge challenge to acquire enough modalities in clinical practice. In this work, a cross-modality reconstruction model, called fine-grain aware generative adversarial network (FA-GAN), is proposed to reconstruct the target modality images of brain from the 2D source modality images with a dual-stages manner. The FA-GAN is able to mine the multi-level shared latent representations from the source modality images and then reconstruct the target modality image from coarse to fine progressively. Specifically, in the coarse stage, the Multi-Grain Extractor firstly extracts and disentangles the shared latent features from the source modality images, and synthesizes the coarse target modality images with a pyramidal network. The Feature-Joint Encoder then encodes the latent features and frequency features jointly. In the fine stage, the Fine-Texture Generator is fed with the joint codes to fine tune the reconstruction of the fine-grained target modality. The wavelet transformation module is employed to extract the frequency codes and guide the Fine-Texture Generator to synthesize finer textures. Comprehensive experiments from MR to PET images on ADNI datasets demonstrate that the proposed model achieves finer structure recovery and outperforms the competing methods quantitatively and qualitatively.

This study focuses on optimizing resource monitoring and management (RMM) of artificial intelligence (AI) systems. The study begins by discussing the fundamental concepts of Digital Twins and cloud collaboration platforms. Subsequently, a collaborative platform called the distributed Digital Twins cloud is designed, incorporating blockchain technology (BCT) to enhance its security. The utilization of cryptographic algorithms in the BCT ensures data security. The study then presents a comprehensive evaluation of the designed BCT-based distributed Digital Twins cloud. The evaluation results demonstrate that the proposed distributed Digital Twins cloud achieves job scheduling times ranging from approximately 35ms to 40ms during training, with the shortest and longest durations observed. The computed data consistency by the proposed platform ranges from 90% to 98%. In the test set, the proposed model achieves job allocation times ranging from approximately 30ms to 35ms, with the shortest and longest durations observed. The data consistency computed by the platform ranges from approximately 90% to 98%, with the lowest and highest values. The network resource management technology model developed in this study exhibits higher efficiency and intelligence, and its performance is comprehensively evaluated. The research findings indicate that the designed model enhances the data processing capability of the cloud platform and effectively improves the security of platform data management, providing substantial support for the advancement of Digital Twins platforms. Moreover, this study offers technical support for the application, design, and optimization of AI technology in network RMM, providing insights into the efficient application and innovation of AI and network technology for future societal development. It contributes to the profound development of AI technology.

With the popularization of the metaverse, virtual reality mapping technology based on digital twins has generated a large amount of spatial data. These data are multidimensional, multi-scale, mobile, and distributed. In order to fully utilize these data, we propose a non mutation multidimensional hash geocoding that can organize and store data with geographic features, and achieve data mapping at different scales from macro to micro. The mapping between them can achieve joint utilization of data of various scales. On this basis, we propose a block network secure storage mapping model for spatial digital twins, which can securely and reliably organize and map spatial data. This article also looks forward to the possible emergence of digital twins of different dimensions and scales in the future metaverse, and proposes an adaptive 3D reconstruction method based on this to adapt to digital twins models of different scales in the metaverse. On the basis of our work, we will further promote the development of the spatial digital twin metaverse.

Gaze point estimation is an essential technology in extended reality devices, and plays a vital role in the metaverse with consumer health. A challenging issue is the great variation underlying the gaze-related encode information, which makes current methods difficult to generate gaze point with stable quality among different persons. To this end, we propose a robust framework to generate high-accuracy gaze point from appearance. Specifically, eyes and face regions are fused into network to enrich high-level features. Then, we develop a common mode features suppression network (CMFS-Net) to predict the gaze bias between the input image and standard image. It is based on the differential amplifier. In addition, a Point-Mean algorithm is designed to generate the estimated gaze point from candidate points. The performance of the CMFS-Net is evaluated on three gaze estimation datasets, GazePC, GazeCapture, and Rice TabletGaze. Among them, the GazePC is collected by ourselves which composes 41.25K images from 165 participants. Experimental results demonstrate that our model is effective and robust for appearance-based gaze point estimation, and has advantages over peer methods.

In the burgeoning metaverse for consumer health (MCH), medical image segmentation methods with high accuracy and generalization capability are essential to drive personalized healthcare solutions and enhance the patient experience. To address the inherent challenges of capturing complex structures and features in medical image segmentation, we propose a convolutional neural network (CNN) and multi-layer-perceptron (MLP) mixed module named HCMM, which hierarchically incorporates local priors of CNN into fully-connected (FC) layers, ingeniously capturing specific details and a broader range of contextual information of the focused object from diverse perspectives. Then, we propose an MLP-based information fusion module (MIF) designed to dynamically merge feature maps of varying levels from different pathways, enhancing feature expression and discriminative power. Based on the above-proposed modules, we design a novel segmentation model, HCMMNet, which can adeptly capture feature maps from input medical images at different scales and perspectives. Through comparative experiments, we demonstrate the outstanding performance of the HCMMNet for medical image segmentation on three publicly available datasets and one self-organized dataset. Notably, our HCMMNet showcases remarkable efficacy while maintaining an extraordinarily lightweight profile, weighing in at a mere 3M, rendering it ideal for MCH application.

Image matching is one of the hottest research topics in the field of computer vision, and it aims to compute the correct feature correspondences between images; however, under the influence of illumination changes and scale changes, relying only on the similarity between feature descriptors as the basis for judging feature matching tends to generate more false feature correspondences. To solve the above problems, in this article, we systematically analyze various factors that lead to false feature correspondences, design corresponding solutions for these various factors, and then propose a multilayer verification (MLV) image-matching method. First, eliminating the potential mismatches caused by brute force matching; second, eliminating the mismatches caused by the ambiguity of descriptors; and third, eliminating the feature matches that do not satisfy geometric constraints. Finally, we test the proposed MLV method on several open-source datasets and make a comprehensive comparison with the classical methods. Experimental results show that the MLV method not only can compute more true feature correspondences but also has a low computational cost. In addition, we applied the MLV method to sparse point cloud reconstruction to demonstrate its practical application value. Source code: https://github.com/caomw/mlv.

Currently, people are actively seeking new and clean energy sources to replace traditional fossil fuels. As a widely distributed, abundant, and easily accessible renewable energy source, wind energy has multiple applications both domestically and internationally, and has a promising future. In wind power forecasting, this paper proposes a new algorithm LFTformer that combines the Transformer model with linear fuzzy information granulation (LFIG).The model uses the improved LFIG algorithm to extract the semantic information of time series, and divides the original time series into multiple information granules, which are then used as inputs to the Transformer model. Through comparative experiments for wind power prediction over 72 hours, 120 hours, and 168 hours, this research demonstrates that the LFTformer can improve the accuracy of wind power prediction. Finally, a Digital Twins system for wind power prediction is constructed using three-dimensional visualization techniques, displaying the prediction results and key data indicators, providing reliable data support for wind power systems.

A high-speed data structure is researched to solve the problem that existing data structure can not support the multi-scale representation of landscape map data on ubiquitous WebGIS. The necessity of the index is analyzed, the algorithm presented, the possibility of supporting landscape map disscussed. In this index, the main tree is obtained from the deformation of the index structure of region quadtree partitioned on the basis of the rule of pyramid structure; a sub-tree structure supports the overlap of landscape map data. The index reflects the changes in spatial resolution according to the depth of the tree. All the nodes of the tree are containers of spatial object. For evaluation, comparative experiments are implemented using our index structure, the result shows that this index method can represent massive landscape map data effectively in WebGIS. The structure has been used in city landscape map on WebGIS and in campus landscape map on mobile phone.