The next generation of mobile communication enables the realization of emerging technologies such as smart cities, smart industries and Internet of vehicles. However, the realization of these technologies requires the mobile communication network to have sensing capability of high precision. Starting from the model of integrated sensing and communication (ISAC) system, the processing of wireless signal sending and receiving and frequency domain radar are discussed, including interpolation, target detection, distance and velocity estimation. Then, according to the characteristics of integrated sensing and communication technology, the application scenarios of ISAC are explored. Next, starting from the application scenario requirements of ISAC, the key indicators of sensing are analyzed, and the performance of sensing capabilities is systematically depicted. Finally, based on 5G-A millimeter wave, the remote detection and accurate tracking capabilities of integrated sensing and communication technology in the low-altitude UAV scene are tested and verified. Through the on-site test, technical support is provided for future integrated communication-sensing applications, and the development of communication technology is promoted towards a more efficient and intelligent direction.

The inherent ambiguity in distinguishing boundaries between lesions and adjacent tissues makes accurate segmentation in ultrasound images challenging. Although deep learning has improved segmentation accuracy, boundary segmentation quality and its relationship with anatomical structures remain underexplored. To address this, we propose DBF-Net, a dual-branch deep neural network that captures supervised relationships between anatomical structures and boundaries. Additionally, we introduce a feature fusion module to enhance the integration of body and boundary information. We evaluate DBF-Net on three public ultrasound image datasets and demonstrate its superiority over existing methods, achieving Dice Similarity Coefficients of 81.05±10.44% for breast cancer, 76.41±5.52% for brachial plexus nerves, and 87.75±4.18% for infantile hemangiomas on the BUSI, UNS, and UHES datasets, respectively. Our approach outperforms current methods, showcasing its effectiveness in ultrasound image segmentation. Code available at: https://github.com/apple1986/DBF-Net.

In convolutional neural networks (CNNs), downsampling operations are crucial to model performance. Although traditional downsampling methods (such as max pooling and strided convolution) perform well in feature aggregation, receptive field expansion, and computational reduction, they may lead to the loss of key spatial information in semantic segmentation tasks, thereby affecting the pixel-by-pixel prediction accuracy. To this end, this study proposes a downsampling method based on information complementarity - Hybrid Pooling Downsampling (HPD). The core is to replace the traditional method with MinMaxPooling, and effectively retain the light and dark contrast and detail features of the image by extracting the maximum value information of the local area. Experiment on various CNN and Transformer architectures using the ACDC and Synapse datasets demonstrate that HPD outperforms traditional segmentation methods. Specifically HPD improves the mean Dice similarity coefficient by 0.5%. The results show that the HPD module provides an efficient solution for semantic segmentation tasks. Code is available at https://github.com/apple1986/HPD.

While patient-specific seizure prediction deep learning (DL) models can deliver remarkable performance tailored to individual patients, the development of patient-independent models that offer satisfactory cross-subject performance holds greater significance and practicality. However, these patient-independent models, which leverage electroencephalogram (EEG) data from multiple patients, may give rise to privacy concerns. This is because EEG data contains sensitive information regarding individuals’ health and mental states. Consequently, from a privacy-preserving perspective, patients may desire the removal of their data information from trained models. Yet, accommodating such forgetting requests presents a formidable challenge: how to enable DL models to forget the data information of specific patients without compromising the performance for others. Although retraining a model from scratch without the data of a specific patient can somewhat address this issue, it becomes computationally prohibitive, especially with large datasets. To tackle this, we introduce an efficient machine unlearning approach called feature unlearning (FU) for seizure prediction. This method modifies the feature projection distribution of specific patients’ data within trained models to match that of models retrained from scratch. Our proposed FU method comprises two primary components: 1) feature shifting, which alters the original distribution of feature projection of specific patients in trained models and feature retaining, which mitigates the adverse effects of feature shifting on other patients, preserving their overall performance through knowledge distillation. We assess our FU method using the CHB-MIT dataset. The results demonstrate that our FU approach can effectively remove the data information of specific patients from trained DL models while maintaining the performance for other patients.

Purpose: Prostate cancer (PCa) risk stratification is of critical importance for clinical diagnosis and treatment planning. Magnetic resonance imaging (MRI) is commonly used for the assessment of PCa risk in patients prior to surgical intervention. However, a significant challenge in PCa risk stratification using MRI is the frequent occurrence of missing MRI modalities. Inferring missing MRI modalities from available data is a crucial step in accurately diagnosing and assessing patient risk. Techniques for synthesizing missing MRI modalities rely extensively on Generative Adversarial Networks (GAN) and their variants. Although GAN is an effective method, the training process often necessitates meticulous tuning of hyperparameters and regularization terms to circumvent issues such as gradient vanishing and mode collapse. In order to overcome these challenges, we propose a generative algorithm based on a Latent Diffusion Model for prostate T2W synthesis based on DWI images. Methods: The algorithm combines a conditional modulation mechanism with a Haar wavelet downsampling strategy to generate specific MRI modalities. The efficacy of the algorithm was evaluated in a comprehensive manner on both private and PI-CAI datasets, with the results demonstrating that it outperforms several popular generative methods, including GAN. Furthermore, the risk stratification of PCa patients was assessed using five deep learning classification models with real and generated MRI images as inputs, respectively. Results: The findings indicated that there was no statistically significant difference in classifying PCa risk stratification when the real or generated MRI images were involved. Conclusion: This approach has the potential to facilitate more accurate PCa risk stratification predictions and personalized treatment for patients.

Privacy protection has become increasingly crucial in the field of epilepsy prediction. Some latest studies introduced the source free domain adaptation (SFDA), which only utilizes a pre-trained source model for protecting the source data privacy. However, the existing SFDA methods exist two shortcomings. (1) the offline setting, which is not suitable for real-world online scenarios (2) the poor performance, which is attributed to the absence of labeled calibration data during the adaptation phase. To this end, we proposed a online seizure prediction framework based on fine-tuning and test-time adaptation (FT3A). Specifically, FT3A employs one seizure event target data to fine-tune and continuously adapt pre-trained source model to unlabeled target data stream. In addition, the adaption and prediction is performed simultaneously. On the one hand, we design the task model as a multi-head structure to increase the confidence of the model and reduce error accumulation. On the other hand, a memory bank is introduced to store a small amount of historical EEG data, which helps handle the catastrophic forgetting concern of the model during online adaptation. Extensive experiments on public CHB-MIT dataset and the private freiburg hospital dataset indicate the superiority and generality of the proposed method.

Prostate cancer (PCa) is a malignancy originating from epithelial cells within the prostate gland. The gold standard for diagnosing PCa is typically based on the Gleason score. However, the inherent variability in biopsy sampling and its potential discordance with radical prostatectomy outcomes can result in the misclassification of the International Society of Urological Pathology (ISUP) Gleason Group (GG). Furthermore, the employment of prostate-specific antigen (PSA) for screening, guiding biopsy decisions, and the reduction in PSA biopsy thresholds has led to a notable increase in unwarranted biopsies among patients with PCa. Consequently, developing an efficient and accurate method to predict ISUP GG is imperative. Currently, most studies focus on ISUP GG binary classification for specific GG. In this study, we leverage multiparametric Magnetic Resonance Imaging (mpMRI) images, encompassing T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and apparent diffusion coefficient (ADC), to introduce a post-fusion based cross-attention model named PFCA-Net. The model is designed to predict ISUP GG categories, specifically GG 0/1, GG 2, GG 3, and GG 4/5. We validate our approach using three distinct deep learning classification models and a mpMRI pre-fusion classification model. Our proposed approach demonstrates exceptional performance, attaining an accuracy (ACC) of 0.9692 and an area under the curve (AUC) of 0.9986, over a dataset comprising 107 PCa patients and encompassing a total of 927 MRI images. This study emphasizes the practical value of mpMRI in distinguishing various ISUP GG categories and expediting the evaluation of PCa GG by clinicians.

Convolutional Neural Networks (CNNs) have been widely used in medical image segmentation to efficiently develop computer-aided diagnosis systems. Due to the locality of convolutional operations, they can be used to extract fine-grained features, but with limitations in building global context and long-range spatial relationships. Recently, shifted window-based multi-layer perceptron (Swin-MLP) methods have demonstrated the ability to learn coarse-grained spatial features in a fixed-size window, but they are not well suited to dense-prediction tasks, such as medical image segmentation. To harmonize the strengths and mitigate the weaknesses of CNNs and SwinMLP in extracting features of varying granularity, we proposed two novel granularity fusion modules that use coarse-grained features to guide the fusion of fine-grained features based on the attention mechanism. Specifically, the first fusion module, named as BGFuse (Block Granularity Fuse), could fuse various scale block-grained features from Swin-MLP. The second fusion module, termed as LGFuse (Local Granularity Fuse), could fuse semantic coarse granularity information into fine granularity features. Equipped with these two fusion modules, we present a new attention-guided encoder-decoder network architecture (termed MGFuseSeg) for medical image segmentation. Without bells and whistles, the proposed MGFuseSeg significantly boosts the performance on three challenging segmentation benchmarks including Synapse, ACDC, and ISIC. Codes are available at https://github.com/apple1986/MGFuseSeg

With AI and big data technologies, telecom operators are looking to change the traditional O&M model from reactive problem handling to proactive prevention and prediction. This paper proposes a model framework trained on multiple data sources for the 5G wireless network to support proactive O&M tasks based on complaint prediction. By grouping user complaints into base station complaint prediction, the model enhanced precision scores while maintaining high recall scores. The model has been integrated into the operator’s work order system to support intelligent operational optimization workflow.

When the data service latency of mobile network is too large, it will cause problems such as slow page opening, game stuck, video stuck and seriously affect user perception. Therefore, optimizing the network and reducing latency become one of the main tasks in mobile network. This paper researches on the analysis method of 5G data service latency problem. A set of analysis methods, which are for problem demarcation and localization, are provided to support network operation and maintenance personnel in improving user perception, focusing on the key performance of wireless and core network networks that affect the service.

We are going to enter the most intelligent era than ever before. Intelligent electronics network is infiltrating into our life and making it more convenient. Nonetheless, users always want smart home be more intelligent and complete more features. users’ issues are endless. Modular packaging device services and effective choreography algorithms can flexible fit different issues. Many organizations have been proving, implementing and managing business solutions for many specific individual industries. However, when comes to smart home for end users, there are numerous limitations in process, tooling, and skills. In the paper, we provide a lightweight visualized service creating tool and an AI-driven service flow construction model. It helps end users to create services though drag-and-drop, and then deploy new services automatically. And in the end a case study will be introduced.

A sharp prediction for ultrasound image segmentation is critical to assist doctors in the accurate diagnosis and treatment. Although the existing methods could achieve impressive performance on biomedical image segmentation, they still have difficulty in segmenting pixels near the boundaries of lesions faithfully. To this end, we propose a body and boundary supervision network aiming to improve ultrasound image segmentation performance, especially on the boundary of the segmented objects. In contrast to existing approaches that take lesions as a whole, a novel dual-branch supervision block was developed to explore explicit modeling of the body and boundary of the object at feature level. Furthermore, we presented an adaptive feature fusion strategy to combine feature maps from dual-branch to get more representative features for final segmentation. Experimental results on three public ultrasound datasets including BUSI, HC18 and Hemangioma demonstrate that the proposed network can achieve leading segmentation performance for breast cancer, fetal head circumference and hemangioma on ultrasound images compared with the existing state-of-the-art models.

With the continuous expansion of network construction, 4G/5G networks have gradually developed into hybrid multi-frequency heterogeneous networks, while the difficulty of inter-RAT mobility assurance is gradually increasing. Traditional interoperability optimization requires enormous labor costs, and the accuracy is low. This paper proposes an AI-based collaborative optimization scheme under multi-frequency heterogeneous 4G/5G networks based on the XGBoost prediction model and DNN algorithm. It aims to comprehensively improve the performance of different users in multi-frequency heterogeneous 4G/5G networks in terms of 4G/5G neighborhood re-organization and intelligent optimization of 4G/5G interoperability parameters. The results show that the proposed scheme has high accuracy and strong generalization, which is critical in improving user mobility perception under complex network structures. The scheme contributes to the network operators’ efficiency improvement and intelligent transformation process.

Semantic segmentation plays crucial role in image analysis, which needs rich spatial information and contextual representations. Recently, deep convolution neural networks have achieved much progress in semantic segmentation. However, it still faces many challenges due to the complexity of images, like the blur, noise, and similarity, which imped the progress of the image segmentation. In this paper, we rethink the feature fusion attention methods, such as spatial attention and channel attention, and propose a novel Fully Attention Module (named FAM) based on every pixel of feature map without reducing the size of the input feature maps. We integrate the proposed FAM into U-Net and Fast-SCNN to assess its effectiveness on Synapse dataset. The extensive experiments show that FAM could improve the performance of both two architectures with acceptable cost in terms of speed and total number of parameters. Specifically, we improve the average Dice Similarity Coefficient (DSC) 4.31% / 1.07% and average Hausdorff Distance (HD) 17.62mm / 8.21mm comparing to U-Net and FastSCNN, respectively. In summary, our proposed FAM could boost the segmentation performance by extracting the fully attention of each pixel in the feature maps. The code will be made publicly available at https://github.com/apple1986/FAM.

The bit error (BER) performance of CDMA in a frequency-selective fading channel can be significantly improved by the use of frequency-domain equalization (FDE). There are three methods can be used to implement FDE in CDMA uplink [I. Martoyo et al., 2003]: the cyclic prefix [K.L. Baum et al., 2002], the zero padding and the overlap-cut (OC). However, the first two methods not only yield lower bandwidth efficiency, but also create higher latency, so the OC is the better choice. In this case, we don't have to modify the structure of the frame. This is very important and useful for applications in which the frame structure is predefined. Thus, this paper proposes an approach to use the improved overlap-cut (IOC) method and parallel interference cancellation (PIC) architecture integrate with frequency domain decision feedback equalization (FD-DFE).

A Code Division Multiple Access (CDMA) downlink receiver based on Adaptive Overlap-Cut (AOC) Method Frequency Domain Equalization (FDE) and non-linear feedback Parallel Interference Cancellation (PIC) is proposed. Firstly, the analysis for the generation and the distribution of the inherent error of Overlap-Cut (OC) Method FDE carries out the improved version of it called AOC Method FDE. Secondly, one PIC scheme with non-linear feedback is designed. After the nonlinear processing, the users' feedback information during the PIC iteration will be adjusted according to its reliability. As a result of both theoretical analysis and simulation, AOC Method FDE outperforms the original OC Method with lower complexity, and the non-linear feedback PIC improves the performance of the receiver further.

In this paper, we propose a hybrid receiver scheme called FDE-PPIC, which combines minimum-mean-square-error frequency domain equalization (MMSE-FDE) with multistage partial parallel interference cancellation (P-PIC), for the downlink of single-carrier cyclic-prefix code division multiple access (CP-CDMA) system over frequency-selective channels. MMSE-FDE instead of conventional RAKE receiver is utilized to alleviate the effect of frequency- selective channel and obtain the initial data estimation. Multistage P-PIC in frequency domain is then employed to mitigate the residual multiuser interference (MUI) for further improving the system performance. Simulation results show that the proposed receiver scheme can achieve better bit error rate (BER) performance than MMSE-FDE and FDE-PIC which is a combination of MMSE-FDE and multistage PIC.

In this paper, we propose a modification of overlap-cut method (OC method) frequency domain equalization (FDE) to reduce complexity in code division multiple access (CDMA) downlink systems. FDE based on fast Fourier transform/inverse fast Fourier transform (FFT/IFFT) carries out single-tap equalization on received signal, moreover, its complexity is much lower than its time-domain counterpart. The analysis for the generation and the distribution of the inherent error of OC method FDE indicates some improvement to the equalizer. As a result of both theoretical analysis and simulation, we obtain a better performance with lower computation burden than conventional algorithms.

The broadcast and multicast service will play a major role in the future wireless communication. The enhanced broadcast and multicast system (EBCMCS) is the evolution of cdma2000 EV/DO in the broadcast system, and can provide the point-to-multipoint broadcast service. In this paper, the quadrature PN spreading and de-spreading methods in the EBCMCS of cdma2000 EV-DO are investigated, and a quadrature PN de-spreading method with lower complexity is proposed. Also, its complexity is compared with the traditional method. Finally, it is proved that the proposed method has much lower complexity than the conventional one without degrading the effectiveness.

Based on the m-sequence, a time-domain channel estimation algorithm for orthogonal frequency division multiplexing (OFDM) is designed. The m-sequence with cyclic prefix (CP) is inserted between OFDM symbols in time domain as training sequence. The cross-correlation is calculated between the received sequence with CP removed and the appointed circular shifted m-sequence, which takes reliability of the estimation of channel impulse response on account of the two-valued auto-correlation property of the m-sequence, thereby the channel frequency response can be advisably decided. As a result of both theoretical analysis and simulation, we obtain an impressive performance with less system overhead and lower computation burden than conventional algorithms.