Fetal congenital heart disease (FCHD) is a common, serious birth defect affecting ∼1% of newborns annually. Fetal echocardiography is the most effective and important technique for prenatal FCHD diagnosis. The prerequisites for accurate ultrasound FCHD diagnosis are accurate view recognition and high-quality diagnostic view extraction. However, these manual clinical procedures have drawbacks such as, varying technical capabilities and inefficiency. Therefore, the automatic identification of high-quality multiview fetal heart scan images is highly desirable to improve prenatal diagnosis efficiency and accuracy of FCHD. Here, we present a framework for multiview fetal heart ultrasound image recognition and quality assessment that comprises two parts: a multiview classification and localization network (MCLN) and an improved contrastive learning network (ICLN). In the MCLN, a multihead enhanced self-attention mechanism is applied to construct the classification network and identify six accurate and interpretable views of the fetal heart. In the ICLN, anatomical structure standardization and image clarity are considered. With contrastive learning, the absolute loss, feature relative loss and predicted value relative loss are combined to achieve favorable quality assessment results. Experiments show that the MCLN outperforms other state-of-the-art networks by 1.52-13.61% when determining the F1 score in six standard view recognition tasks, and the ICLN is comparable to the performance of expert cardiologists in the quality assessment of fetal heart ultrasound images, reaching 97% on a test set within 2 points for the four-chamber view task. Thus, our architecture offers great potential in helping cardiologists improve quality control for fetal echocardiographic images in clinical practice.

Abstract Segmenting heart components in the apical four‐chamber view of fetal echocardiography is of critical significance in clinical practice. However, it is difficult to recognize these components due to small‐scale components and the imbalanced ventricular apex orientation. In this study, a novel segmentation framework is proposed to segment ten general fetal heart components for the first time. This framework consists of a multi‐directional fine‐density (MDFD) data augmentation method and a coarse‐to‐fine cascade network (CFCN). MDFD enhances the apex orientation diversity and balances the orientation distribution. CFCN has two stages including a coarse network and a fine network. These two stages have similar structures that consist of a feature extractor and a feature refined layer named as Element‐Wise Power with Dynamic Exponent layer (EWPDE). EWPDE which is a plug‐and‐play module for segmentation refines the features from the feature extractor to position small components accurately. By adopting EWPDE, the influence of each pixel is adjusted and hard pixels of small components are segmented precisely. Based on the dataset, the method is proved to be effective with the high mean intersection over union (mIoU) value and low missing ratio (MR). With MDFD and EWPDE, CFCN that adopts DeepLabV3+ as the feature extractor outperforms the best segmentation results (mIoU:0.480, MR:0.035). Compared to the original performance (mIoU:0.407, MR:0.085) of DeepLabV3+, the method improves the results significantly.

Fetal congenital heart disease (FHD) is a common and serious congenital malformation in children. In Asia, FHD birth defect rates have reached as high as 9.3%. For the early detection of birth defects and mortality, echocardiography remains the most effective method for screening fetal heart malformations. However, standard echocardiograms of the fetal heart, especially four-chamber view images, are difficult to obtain. In addition, the pathophysiological changes in fetal hearts during different pregnancy periods lead to ever-changing two-dimensional fetal heart structures and hemodynamics, and it requires extensive professional knowledge to recognize and judge disease development. Thus, research on the automatic screening for FHD is necessary. In this paper, we proposed a new model named DGACNN that shows the best performance in recognizing FHD, achieving a rate of 85%. The motivation for this network is to deal with the problem that there are insufficient training datasets to train a robust model. There are many unlabeled video slices, but they are tough and time-consuming to annotate. Thus, how to use these un-annotated video slices to improve the DGACNN capability for recognizing FHD, in terms of both recognition accuracy and robustness, is very meaningful for FHD screening. The architecture of DGACNN comprises two parts, that is, DANomaly and GACNN (Wgan-GP and CNN). DANomaly, similar to the ALOCC network, but incorporates cycle adversarial learning to train an end-to-end one-class classification (OCC) network that is more robust and has a higher accuracy than ALOCC in screening video slices. For the GACNN architecture, we use FCH (four chamber heart) video slices at around the end-systole, as screened by DANomaly, to train a WGAN-GP for the purpose of obtaining ideal low-level features that can robustly improve the FHD recognition accuracy. A few annotated video slices, as screened by DANomaly, can also be used for data augmentation so as to improve the FHD recognition further. The experiments show that the DGACNN outperforms other state-of-the-art networks by 1%-20% in recognizing FHD. A comparison experiment shows that the proposed network already outperforms the performance of expert cardiologists in recognizing FHD, reaching 84% in a test. Thus, the proposed architecture has high potential for helping cardiologists complete early FHD screenings.

What topologies should be used to evaluate protocols for interdomain routing? Using the most current Internet topology is not practical since its size is prohibitive for detailed, packet-level interdomain simulations. Besides being of moderate size, the topology should be policy-aware, that is, it needs to represent business relationships between adjacent nodes (that represent autonomous systems). In this paper, we address this issue by providing a framework to generate small, realistic, and policy-aware topologies. We propose HBR, a novel sampling method, which exploits the inherent hierarchy of the policy-aware Internet topology. We formally prove that our approach generates connected and legitimate topologies, which are compatible with the policy-based routing conventions and rules. Using simulations, we show that HBR generates topologies that: 1) maintain the graph properties of the real topology; 2) provide reasonably realistic interdomain simulation results while reducing the computational complexity by several orders of magnitude as compared to the initial topology. Our approach provides a permanent solution to the problem of interdomain routing evaluations: Given a more accurate and complete topology, HBR can generate better small topologies in the future.

The Internet community has not reached a consensus on an appropriate topological model for evaluating the performance of inter-domain routing protocols. Using the current Internet topology is not realistic, since its size is prohibitively large for, say, a packet-level BGP simulation. Furthermore, routing policies, which play a critical role in inter-domain routing, are often ignored in many simulation studies. In this paper, we address this issue by designing an algorithm to generate small-scale, realistic, and policy-aware topologies. We propose HBR, a network sampling method, which produces topologies that preserve the fundamental properties of the Internet graph, including, in particular, its hierarchical structure. Our approach provides a long-term solution to the difficult problem of AS-level routing evaluations: it can be used to generate small realistic topologies in the future, starting from any newer or more complete Internet instance.

The bandwidth of multiple physical communication links can be aggregated with inverse multiplexing to create a single, higher capacity logical communication link. However, achieving the maximum possible aggregated bandwidth becomes challenging as the communication characteristics of the underlying links (e.g., available bandwidth, packet loss rate) change with time. In this paper we introduce NATALIE, a Network-aware traffic equalizer. NATALIE combines an arbitrary set of network interfaces and schedules IP packet transmissions over those interfaces in a weighted round-robin fashion. To achieve the maximum possible throughput, NATALIE measures the communication characteristics of the underlying links and dynamically adjusts its scheduler to assign packets to each link in proportion to its available capacity. We describe NATALIE'S implementation, the test environment we built on the Emulab Network Testbed, and the results of experiments demonstrating NATALIE'S throughput performance for TCP connections over diverse underlying links with cross-traffic.

Routing asymmetry in the Internet can significantly affect the manner in which we model and simulate its behavior. In this paper, we study routing asymmetry in the Internet and present quantitative evaluations on the extent of such asymmetry today. Our quantitative evaluations provide a measure of the difference between the forward and reverse paths between two end points. Routing asymmetry has not been studied extensively before; this is primarily due to the lack of a systematic approach for quantifying asymmetry except for simply computing the difference between the forward and reverse path lengths. By applying our framework for representing asymmetry, we quantify routing asymmetry for both US higher education academic networks and general commercial networks at two different levels: the autonomous system (AS) level and the router (or link) level. We take into consideration, not only the difference in the forward and reverse path lengths, but also the AS and link identities and the sequence in which these entities appear on the paths. We measure the AS level routing asymmetry, and provide upper lower bounds on link level routing asymmetry. Our studies show that academic networks appear to be more symmetric than general commercially deployed networks. Furthermore, our studies demonstrate that routing asymmetry exhibits a skewed distribution i.e., a few end-points seem to display a higher extent of participation on asymmetric routes.

Our objective is to quantify the extent of routing asymmetry in the Internet: the measure of the difference between the forward and backward paths between two end points. Routing asymmetry has not been studied extensively. Most previous studies only consider asymmetry in terms of length and there is a lack of a systematic approach for quantifying asymmetry. One of the challenges in quantifying asymmetry is the formulation of an appropriate set of metrics that can effectively capture various notions of asymmetry. We point out that asymmetry could be of various types. We propose a framework to quantify the routing asymmetry between end hosts and propose two new metrics: absolute asymmetry and length-based normalized asymmetry. Our metrics capture the differences in router/AS identities, their appearing sequences and path lengths in a seamless way. We apply our framework to real Internet measurement data and examine routing asymmetry at the autonomous system (AS) level. We deduce the routing asymmetry distribution based on our framework, and we find that about 14% of pairs of routes considered display AS level routing asymmetry. Furthermore, our studies demonstrate that the routing asymmetry exhibits a skewed distribution, since a few end-points are consistently members of asymmetric pairs.

Interception of a data stream is central to any intelligent and dynamic processing of Web information. It is perhaps as fundamental to Internet services' overall architecture as the design of disk scheduling to the conventional machine architecture. We discuss an IPv6 based indexing protocol that can facilitate random access into multilevel hierarchically encoded content streams and provide serious performance boost to Web content processing.