Search

Abstract Autism Spectrum Disorder (ASD) is a prevalent neurological condition with multiple co-occurring comorbidities that seriously affect mental health. Precisely diagnosis of ASD is crucial to intervention and rehabilitation. A single modality may not fully reflect the complex mechanisms underlying ASD, and combining multiple modalities enables a more comprehensive understanding. Here, we propose, DeepASD, an end-to-end trainable regularized graph learning method for ASD prediction, which incorporates heterogeneous multimodal data and latent inter-patient relationships to better understand the pathogenesis of ASD. DeepASD first learns cross-modal feature representations through a multimodal adversarial-regularized encoder, and then constructs adaptive patient similarity networks by leveraging the representations of each modality. DeepASD exploits inter-patient relationships to boost the ASD diagnosis that is implemented by a classifier compositing of graph neural networks. We apply DeepASD to the benchmarking Autism Brain Imaging Data Exchange (ABIDE) data with four modalities. Experimental results show that the proposed DeepASD outperforms eight state-of-the-art baselines on the benchmarking ABIDE data, showing an improvement of 13.25% in accuracy, 7.69% in AUC-ROC, and 17.10% in specificity. DeepASD holds promise for a more comprehensive insight of the complex mechanisms of ASD, leading to improved diagnosis performance.

Summary: Accurate detection of liver lesions from multi-phase contrast-enhanced CT (CECT) scans is a fundamental step for precise liver diagnosis and treatment. However, the analysis of multi-phase contexts is heavily challenged by the misalignment caused by respiration coupled with the movement of organs. Here, we proposed an AI system for multi-phase liver lesion segmentation (named MULLET) for precise and fully automatic segmentation of real-patient CECT images. MULLET enables effectively embedding the important ROIs of CECT images and exploring multi-phase contexts by introducing a transformer-based attention mechanism. Evaluated on 1,229 CECT scans from 1,197 patients, MULLET demonstrated significant performance gains in terms of Dice, Recall, and F2 score, which are 5.80%, 6.57%, and 5.87% higher than state of the arts, respectively. MULLET has been successfully deployed in real-world settings. The deployed AI web server provides a powerful system to boost clinical workflows of liver lesion diagnosis and could be straightforwardly extended to general CECT analyses.

Academic knowledge graphs are essential resources and can be beneficial in widespread real-world applications. Most of the existing academic knowledge graphs are far from completion; thus, knowledge graph completion—the task of extending a knowledge graph with missing entities and relations—attracts many researchers. Most existing methods utilize low-dimensional embeddings to represent entities and relations and follow the discrimination paradigm for link prediction. However, discrimination approaches may suffer from the scaling issue during inference with large-scale academic knowledge graphs. In this paper, we propose a novel approach of a generative transformer with knowledge-guided decoding for academic knowledge graph completion. Specifically, we introduce generative academic knowledge graph pre-training with a transformer. Then, we propose knowledge-guided decoding, which leverages relevant knowledge in the training corpus as guidance for help. We conducted experiments on benchmark datasets for knowledge graph completion. The experimental results show that the proposed approach can achieve performance gains of 30 units of the MRR score over the baselines on the academic knowledge graph AIDA.

Primary adenoid cystic carcinoma (ACC) of the upper anterior mediastinum mimicking a thyroid tumor has rarely been seen in clinical practice and lacks a standard of care therapy. Here, we report a 47-year old female patient with an ACC originated from the upper anterior mediastinum presenting as a thyroid gland tumor. The patient received gross surgical resection of the tumor and underwent post-surgical chemotherapy and radiotherapy. The patient was free from local recurrence 3-years following initial treatment, but developed multiple lung metastases. She remains under clinical observation without discomfort and is still followed as an outpatient. Here, we also summarized recent reports of similar cases with hope to provide some experience for future clinical practice.

Wireless link correlation has shown significant impact on the performance of various sensor network protocols. Many works have been devoted to exploiting link correlation for protocol improvements. However, the effectiveness of these designs heavily relies on the accuracy of link correlation measurement. In this paper, we investigate state-of-the-art link correlation measurement and analyze the limitations of existing works. We then propose a novel lightweight and accurate link correlation estimation (LACE) approach based on the reasoning of link correlation formation. LACE combines both long-term and short-term link behaviors for link correlation estimation. We implement LACE as a stand-alone interface in TinyOS and incorporate it into both routing and flooding protocols. Simulation and testbed results show that LACE: (1) achieves more accurate and lightweight link correlation measurements than the state-of-the-art work; and (2) greatly improves the performance of protocols exploiting link correlation.

Many trace-based diagnostic techniques have been proposed for abnormal detection and fault diagnosis in networked embedded systems such as wireless sensor networks (WSNs). Event tracing is a nontrivial task for resource-constrained embedded devices. Existing tracing approaches employ compression algorithms to reduce the trace size. However, these approaches either are inapplicable or perform poorly. In this paper, we propose TOC, a novel event tracing technique using online compression. TOC combines periodical pattern mining and efficient token assignment, effectively reducing the trace size with acceptable execution overhead. We implement TOC based on TinyOS 2.1.2 and evaluate its effectiveness by case studies in sensor network applications. Results show that TOC reduces the trace size by 52.2%, compared with LIS—a state-of-the-art event tracing method.

In wired networks, monitor-based network tomography has been proved to be an effective technology for network internal state measurements. Existing wired network tomography approaches assume that the network topology is relatively static. However, the network topology of sensor networks is usually changing over time due to wireless dynamics. In this paper, we study the problem to assign a number of sensor nodes as monitors in large scale sensor networks, so that the end-to-end measurements among monitors can be used to identify hop-by-hop link metrics. We propose RoMA, a Robust Monitor Assignment, algorithm to assign monitors in large scale sensor networks with dynamically changing topology. RoMA includes two components, confidence-based robust topology generation and cost-minimized monitor assignment. We implement RoMA and evaluate its performance based on a deployed large scale sensor network. Results show that RoMA achieves high identifiability with dynamically changing topology and is able to assign monitors with minimum cost.

Large-scale wireless sensor networks follow the two-tiered architecture, where master nodes take charge of storing data and processing queries. However, if a master node is compromised, the information stored in it may be exposed, and query results can be juggled. This paper presents a novel scheme called SEF for secure range queries. To preserve privacy, SEF employs the order-preserving symmetric encryption which not only supports efficient range queries, but also maintains a strong security standard. To preserve authenticity and integrity of query results, we propose a novel data structure called Authenticity & Integrity tree. Moreover, SEF is flexible since it allows users to include or exclude the authenticity and integrity guarantee. To the best of our knowledge, this paper is the first to use the characteristic of NAND flash to achieve high storage utilization and query processing efficiency. The efficiency of the proposed scheme is demonstrated by experiments on real sensor platforms.