Rethinking Pooling Operation for Liver and Liver-Tumor Segmentations

Deep convolutional neural networks (DCNNs) have been widely used in medical image segmentation due to their excellent feature learning ability. In these DCNNs, the pooling operation is usually used for image down-sampling, which can gradually reduce the image resolution and thus expands the receptive field of convolution kernel. Although the pooling operation has the above advantages, it inevitably causes information loss during the down-sampling of the pooling process. This paper proposes an effective weighted pooling operation to address the problem of information loss. First, we set up a pooling window with learnable parameters, and then update these parameters during the training process. Secondly, we use weighted pooling to improve the full-scale skip connection and enhance the multi-scale feature fusion. We evaluated weighted pooling on two public benchmark datasets, the LiTS2017 and the CHAOS. The experimental results show that the proposed weighted pooling operation effectively improve network performance and improve the accuracy of liver and liver-tumor segmentation.

Multi-Frequency Radar Micro-Doppler Based Classification of Micro-Drone Payload Weight

The use of drones for recreational, commercial and military purposes has seen a rapid increase in recent years. The ability of counter-drone detection systems to sense whether a drone is carrying a payload is of strategic importance as this can help determine the potential threat level posed by a detected drone. This paper presents the use of micro-Doppler signatures collected using radar systems operating at three different frequency bands for the classification of carried payload of two different micro-drones performing two different motions. Use of a KNN classifier with six features extracted from micro-Doppler signatures enabled mean payload classification accuracies of 80.95, 72.50 and 86.05%, for data collected at S-band, C-band and W-band, respectively, when the drone type and motion type are unknown. The impact on classification performance of different amounts of situational information is also evaluated in this paper.

Persymmetric Adaptive Union Subspace Detection

This paper addresses the detection of a signal belonging to several possible subspace models, namely, a union of subspaces (UoS), where the active subspace that generated the observed signal is unknown. By incorporating the persymmetric structure of received data, we propose three UoS detectors based on GLRT, Rao, and Wald criteria to alleviate the requirement of training data. In addition, the detection statistic and classification bound for the proposed detectors are derived. Monte-Carlo simulations demonstrate the detection and classification performance of the proposed detectors over the conventional detector in training-limited scenarios.

Interference Utilization Precoding in Multi-Cluster IoT Networks

In Internet-of-Things, downlink multi-device interference has long been considered as a harmful element deteriorating system performance, and thus the principle of the classic interference-mitigation based precoding is to suppress the multi-device interference by exploiting the spatial orthogonality. In recent years, a judicious interference utilization precoding has been developed, which is capable of exploiting multi-device interference as a beneficial element for improving device’s reception performance, thus reducing downlink communication latency. In this review paper, we aim to review the emerging interference utilization precoding techniques. We first briefly introduce the concept of constructive interference, and then we present two generic downlink interference-utilization optimizations, which utilizes the multi-device interference for enhancing system performance. Afterwards, the application of interference utilization precoding is discussed in multi-cluster scenario. Finally, some open challenges and future research topics are envisaged.

Deep Learning-Based Object Tracking via Compressed Domain Residual Frames

The extensive rise of high-definition CCTV camera footage has stimulated both the data compression and the data analysis research fields. The increased awareness of citizens to the vulnerability of their private information, creates a third challenge for the video surveillance community that also has to encompass privacy protection. In this paper, we aim to tackle those needs by proposing a deep learning-based object tracking solution via compressed domain residual frames. The goal is to be able to provide a public and privacy-friendly image representation for data analysis. In this work, we explore a scenario where the tracking is achieved directly on a restricted part of the information extracted from the compressed domain. We utilize exclusively the residual frames already generated by the video compression codec to train and test our network. This very compact representation also acts as an information filter, which limits the amount of private information leakage in a video stream. We manage to show that using residual frames for deep learning-based object tracking can be just as effective as using classical decoded frames. More precisely, the use of residual frames is particularly beneficial in simple video surveillance scenarios with non-overlapping and continuous traffic.

Precoded Cluster Hopping for Multibeam GEO Satellite Communication Systems

Beam hopping (BH) and precoding are two trending technologies for high-throughput satellite (HTS) systems. While BH enables the flexible adaptation of the offered capacity to the heterogeneous demand, precoding aims at boosting the spectral efficiency. In this study, we consider an HTS system that employs BH in conjunction with precoding in an attempt to bring the benefits of both in one. In particular, we propose the concept of cluster hopping (CH), where a set of adjacent beams are simultaneously illuminated with the same frequency resource. On this line, we propose an efficient time–space illumination pattern design, where we determine the set of clusters that shall be illuminated simultaneously at each hopping event along with the dwelling time. The CH time–space illumination pattern design formulation is shown to be theoretically intractable due to the combinatorial nature of the problem and the impact of the actual illumination design on the resulting interference. For this, we make some design decisions on the beam–cluster design that open the door to a less complex still well-performing solution. Supporting results based on numerical simulations are provided which validate the effectiveness of the proposed CH concept and a time–space illumination pattern design with respect to benchmark schemes.

New Measures of Heart Rate Variability Based on Subband Tachogram Complexity and Spectral Characteristics for Improved Stress and Anxiety Monitoring in Highly Ecological Settings

Prediction of mental states, such as stress and anxiety, can be important in situations where reduced job performance due to increased mental strain can lead to critical situations (e.g., front-line healthcare workers and first responders). While recent advances in biomedical wearable sensor technologies have allowed for collection of multiple physiological signals in everyday environments, numerous challenges emerge from such uncontrolled settings, including increased noise levels and artifacts, confounding effects from other psychological states (e.g., mental fatigue), as well as physical variables (e.g., physical activity). These factors can be particularly detrimental for heart rate variability (HRV) measures which, in controlled settings, have been shown to accurately track stress and anxiety states. In this paper, we propose two new ways of computing HRV proxies which we show are more robust to such artifacts and confounding factors. The proposed features measure spectral and complexity properties of different aspects of the autonomic nervous system, as well as their interaction. Across two separate “in-the-wild” datasets, the proposed features showed to not only outperform benchmark HRV metrics, but to also provide complementary information, thus leading to significantly greater accuracy levels when fused together. Feature ranking analysis further showed the proposed features appearing in 45–64% of the top features, thus further emphasizing their importance. In particular, features derived from the high frequency band showed to be most important in the presence of fatigue and physical activity confounding factors, thus corroborating their importance for mental state assessment in highly ecological settings.

EKF- and UKF-Based Estimators for Radar System

In radar, the measurements (like the range and radial velocity) are determined from the time delay and Doppler shift. Since the time delay and Doppler shift are estimated from the phase of the received echo, the concerned estimation problem is nonlinear. Consequently, the conventional estimator based on the fast Fourier transform (FFT) is prone to yield high estimation errors. Recently, nonlinear estimators based on kernel least mean square (KLMS) are introduced and found to outperform the conventional estimator. However, estimators based on KLMS are susceptible to incorrect choice of various system parameters. Thus, to mitigate the limitation of existing estimators, in this paper, two efficient low-complexity nonlinear estimators, namely, the extended Kalman filter (EKF) and the unscented Kalman filter (UKF), are proposed. The EKF is advantageous due to its implementation simplicity; however, it suffers from the poor representation of the nonlinear functions by the first-order linearization, whereas UKF outperforms the EKF and offers better stability due to exact consideration of the system nonlinearity. Simulation results reveal improved accuracy achieved by the proposed EKF- and UKF-based estimators.