Signed graph representation learning for functional-to-structural brain network mapping.

MRI-derived brain networks have been widely used to understand functional and structural interactions among brain regions, and factors that affect them, such as brain development and diseases. Graph mining on brain networks can facilitate the discovery of novel biomarkers for clinical phenotypes and neurodegenerative diseases. Since brain functional and structural networks describe the brain topology from different perspectives, exploring a representation that combines these cross-modality brain networks has significant clinical implications. Most current studies aim to extract a fused representation by projecting the structural network to the functional counterpart. Since the functional network is dynamic and the structural network is static, mapping a static object to a dynamic object may not be optimal. However, mapping in the opposite direction (i.e., from functional to structural networks) are suffered from the challenges introduced by negative links within signed graphs. Here, we propose a novel graph learning framework, named as Deep Signed Brain Graph Mining or DSBGM, with a signed graph encoder that, from an opposite perspective, learns the cross-modality representations by projecting the functional network to the structural counterpart. We validate our framework on clinical phenotype and neurodegenerative disease prediction tasks using two independent, publicly available datasets (HCP and OASIS). Our experimental results clearly demonstrate the advantages of our model compared to several state-of-the-art methods.

Contrastive Brain Network Learning via Hierarchical Signed Graph Pooling Model.

Recently, brain networks have been widely adopted to study brain dynamics, brain development, and brain diseases. Graph representation learning techniques on brain functional networks can facilitate the discovery of novel biomarkers for clinical phenotypes and neurodegenerative diseases. However, current graph learning techniques have several issues on brain network mining. First, most current graph learning models are designed for unsigned graph, which hinders the analysis of many signed network data (e.g., brain functional networks). Meanwhile, the insufficiency of brain network data limits the model performance on clinical phenotypes' predictions. Moreover, few of the current graph learning models are interpretable, which may not be capable of providing biological insights for model outcomes. Here, we propose an interpretable hierarchical signed graph representation learning (HSGPL) model to extract graph-level representations from brain functional networks, which can be used for different prediction tasks. To further improve the model performance, we also propose a new strategy to augment functional brain network data for contrastive learning. We evaluate this framework on different classification and regression tasks using data from human connectome project (HCP) and open access series of imaging studies (OASIS). Our results from extensive experiments demonstrate the superiority of the proposed model compared with several state-of-the-art techniques. In addition, we use graph saliency maps, derived from these prediction tasks, to demonstrate detection and interpretation of phenotypic biomarkers.

Case report: Anti-NMDA receptor encephalitis manifesting as rapid weight loss and abnormal movement disorders with alternating unilateral ptosis and contralateral limb tremor.

Anti-N-methyl-D-aspartate receptor (NMDAR) encephalitis, associated with immunoglobulin G (IgG) autoantibodies against the GluN1 subunit of the NMDAR, is one of the most common types of autoimmune encephalitis. In patients with anti-NMDAR encephalitis, movement disorders (MDs) are often frequent, mainly presenting as facial dyskinesias and stereotyped movements. The alternating clinical manifestation of limb tremor with unilateral ptosis is rare. Here, we report an interesting case of a 22-year-old woman with rapid weight loss presenting with staged dyskinesia. Interestingly, she typically showed persistent tremor of the right upper limb, which would stop when her left upper eyelid drooped uncontrollably, a phenomenon that lasted for a few seconds, followed by automatic upper eyelid lift and continued persistent tremor of the upper limb. Moreover, it was fortunate to find anti-NMDAR antibodies in her cerebrospinal fluid (CSF), which indicated the patient had anti-NMDAR encephalitis. And abnormal apparent diffusion coefficient (ADC) hyperintense signals on the left midbrain interpeduncular fossa explained this manifestation of focal neurological deficit. After the systematic administration of immunotherapy (intravenous immunoglobulin, IVIG), steroid pulse therapy, and symptomatic treatment, the initial symptoms were significantly relieved except for limb tremor. The MDs were becoming less visible for the next six months under topiramate prescriptions. Noteworthy, there are no specific MD phenotypes in anti-NMDAR encephalitis. We describe the young women with unique MDs and rapid weight loss to help us get a more comprehensive understanding of anti-NMDAR encephalitis.

Coexistence of anti-SOX1 and anti-GABAB receptor antibodies with paraneoplastic limbic encephalitis presenting with seizures and memory impairment in small cell lung cancer: A case report.

Purpose: Paraneoplastic neurological syndromes associated with autoantibodies are rare diseases that cause abnormal manifestations of the nervous system. Early diagnosis of paraneoplastic neurological syndromes paves the way for prompt and efficient therapy. Case report: we reported a 56-year-old man presenting with seizures and rapidly progressive cognitive impairment diagnosed as paraneoplastic limbic encephalitis (PLE) with anti-SRY-like high-mobility group box-1 (SOX-1) and anti-γ-aminobutyric acid B (GABAB) receptor antibodies and finally confirmed by biopsy as small cell lung cancer (SCLC). At the first admission, brain magnetic resonance imaging (MRI) showed no abnormal signal in bilateral hippocampal regions and no abnormal enhancement of enhanced scan. The serum anti-GABAB receptor antibody was 1:100 and was diagnosed as autoimmune encephalitis (AE). The computed tomography (CT) scans of the chest showed no obvious tumor signs for the first time. Although positron emission tomography-computed tomography (PET-CT) revealed hypermetabolism in the para mid-esophageal, the patient and his family declined to undertake a biopsy. The patient improved after receiving immunoglobulin, antiepileptic therapy, and intravenous methylprednisolone (IVMP) pulse treatment. However, after 4 months, the symptoms reappeared. Brain MRI revealed abnormal signals in the hippocampal regions. Reexamination of the cerebral fluid revealed anti-GABAB receptor and anti-SOX-1 antibodies, which contributed to the diagnosis of PLE. SCLC was found in a para mid-esophageal pathological biopsy. Antiepileptic medications and immunoglobulin were used to treat the patient, and the symptoms were under control. Conclusion: Our findings increase the awareness that patients with limbic encephalitis with cognitive dysfunction and epileptic seizures should be enhanced to detect latent malignancy. Our case also highlights the importance of anti-SOX1 antibodies in the detection of underlying neoplasm, particularly SCLC. Our findings raise awareness of the cognitive impairment seen by patients with limbic encephalitis.

A Hierarchical Graph Learning Model for Brain Network Regression Analysis.

Brain networks have attracted increasing attention due to the potential to better characterize brain dynamics and abnormalities in neurological and psychiatric conditions. Recent years have witnessed enormous successes in deep learning. Many AI algorithms, especially graph learning methods, have been proposed to analyze brain networks. An important issue for existing graph learning methods is that those models are not typically easy to interpret. In this study, we proposed an interpretable graph learning model for brain network regression analysis. We applied this new framework on the subjects from Human Connectome Project (HCP) for predicting multiple Adult Self-Report (ASR) scores. We also use one of the ASR scores as the example to demonstrate how to identify sex differences in the regression process using our model. In comparison with other state-of-the-art methods, our results clearly demonstrate the superiority of our new model in effectiveness, fairness, and transparency.

White Matter Alterations in Depressive Disorder.

Depressive disorder is the most prevalent affective disorder today. Depressive disorder has been linked to changes in the white matter. White matter changes in depressive disorder could be a result of impaired cerebral blood flow (CBF) and CBF self-regulation, impaired blood-brain barrier function, inflammatory factors, genes and environmental factors. Additionally, white matter changes in patients with depression are associated with clinical variables such as differential diagnosis, severity, treatment effect, and efficacy assessment. This review discusses the characteristics, possible mechanisms, clinical relevance, and potential treatment of white matter alterations caused by depressive disorders.

MXene/ZIF-67/PAN Nanofiber Film for Ultra-sensitive Pressure Sensors.

Flexible pressure sensors may be used in electronic skin (e-skin), artificial intelligence devices, and disease diagnosis, which require a large response range and high sensitivity. An appropriate design of the structure of the active layer can help effectively solve this problem. Herein, we aim at developing a wearable pressure sensor using the MXene/ZIF-67/polyacrylonitrile (PAN) nanofiber film, fabricated by electrospinning technology. Owing to the rough structure and three-dimensional network architecture, the MXene/ZIF-67/PAN film-based device displays a broad working range (0-100 kPa), good sensitivity (62.8 kPa-1), robust mechanical stability (over 10,000 cycles), and fast response/recovery time (10/8 ms). Moreover, the fabricated pressure sensors can be used to detect and differentiate between different body motion information, including elbow bending, finger movements, and wrist pulses. Overall, this design of a rough three-dimensional conductive network structure shows potential in the field of wearable electronics and medical devices.

Cerebral blood flow self-regulation in depression.

BACKGROUND: Depression is a common neuropsychiatric disease with a high prevalence rate. Sleep problems, memory decline, dizziness and headaches are the most common neurological symptoms in depressed patients. Abnormality of cerebral blood flow (CBF) has been observed in depressive patients, but those patients did not have intracranial structural damage. Both of those phenomena might be related to cerebral blood flow self-regulation (CBFSR: cerebral blood flow self-regulation). CBFSR can maintain CBF relatively stable in response to changes in neurological and metabolic factors. Therefore, this review aimed to discuss CBFSR in depression. METHODS: We searched for keywords such as "depression", "cerebral blood flow", "cerebral autoregulation", "cerebrovascular reactivity" and the words related to depression. We analyzed whether there is a change in the CBFSR in depression, further explored whether there is a relationship between the pathogenesis of depression and the CBFSR, and discussed the possible mechanism of impaired CBFSR in patients with depression. RESULTS: Discovered by the literature review, CBFSR is significantly impaired in depressed patients. The level of circulating markers of endothelial dysfunction, nitric oxide, inflammatory cytokines, glucocorticoid and monoamine neurotransmitters is mostly abnormal in depression, which affected the CBFSR to varying degrees. LIMITATIONS: Limitations include the small number of direct studies about depression and CBFSR mechanisms. CONCLUSION: CBFSR is impaired in depression. The underlying mechanisms include endothelial dysfunction, overactivation of microglia and changes of cytokines, hyperactivation of the HPA axis, increased oxidative stress, monoamine neurotransmitter disorders, etc. These deepened our understanding of the clinical symptoms of depressed patients.

HIERARCHICAL BRAIN EMBEDDING USING EXPLAINABLE GRAPH LEARNING.

Brain networks have been extensively studied in neuroscience, to better understand human behavior, and to identify and characterize distributed brain abnormalities in neurological and psychiatric conditions. Several deep graph learning models have been proposed for brain network analysis, yet most current models lack interpretability, which makes it hard to gain any heuristic biological insights into the results. In this paper, we propose a new explainable graph learning model, named hierarchical brain embedding (HBE), to extract brain network representations based on the network community structure, yielding interpretable hierarchical patterns. We apply our new method to predict aggressivity, rule-breaking, and other standardized behavioral scores from functional brain networks derived using ICA from 1,000 young healthy subjects scanned by the Human Connectome Project. Our results show that the proposed HBE outperforms several state-of-the-art graph learning methods in predicting behavioral measures, and demonstrates similar hierarchical brain network patterns associated with clinical symptoms.

3D bi-directional transformer U-Net for medical image segmentation.

As one of the popular deep learning methods, deep convolutional neural networks (DCNNs) have been widely adopted in segmentation tasks and have received positive feedback. However, in segmentation tasks, DCNN-based frameworks are known for their incompetence in dealing with global relations within imaging features. Although several techniques have been proposed to enhance the global reasoning of DCNN, these models are either not able to gain satisfying performances compared with traditional fully-convolutional structures or not capable of utilizing the basic advantages of CNN-based networks (namely the ability of local reasoning). In this study, compared with current attempts to combine FCNs and global reasoning methods, we fully extracted the ability of self-attention by designing a novel attention mechanism for 3D computation and proposed a new segmentation framework (named 3DTU) for three-dimensional medical image segmentation tasks. This new framework processes images in an end-to-end manner and executes 3D computation on both the encoder side (which contains a 3D transformer) and the decoder side (which is based on a 3D DCNN). We tested our framework on two independent datasets that consist of 3D MRI and CT images. Experimental results clearly demonstrate that our method outperforms several state-of-the-art segmentation methods in various metrics.

Knitted Ti3C2Tx MXene based fiber strain sensor for human-computer interaction.

Fiber-based stretchable electronics with feasibility of weaving into textiles and advantages of light-weight, long-term stability, conformability and easy integration are highly desirable for wearable electronics to realize personalized medicine, artificial intelligence and human health monitoring. Herein, a fiber strain sensor is developed based on the Ti3C2Tx MXene wrapped by poly(vinylidenefluoride-co-trifluoroethylene) (P(VDF-TrFE)) polymer nanofibers prepared via electrostatic spinning. Owing to the good conductivity of Ti3C2Tx and unique 3D reticular structure with wave shape, the resistance of Ti3C2Tx@P(VDF-TrFE) polymer nanofibers changes under external force, thus providing remarkable strain inducted sensing performance. As-fabricated sensor exhibits high gauge factor (GF) of 108.8 in range of 45-66% strain, rapid response of 19 ms, and outstanding durability over 1600 stretching/releasing cycles. The strain sensor is able to monitor vigorous human motions (finger or wrist bending) and subtle physiological signals (blinking, pulse or voice recognition) in real-time. Moreover, a data glove is designed to connect different gestures and expressions to form an intelligent gesture-expression control system, further confirming the practicability of our Ti3C2Tx@P(VDF-TrFE) strain sensors in multifunctional wearable electronic devices.

Flexible Self-Powered Integrated Sensing System with 3D Periodic Ordered Black Phosphorus@MXene Thin-Films.

Accurate and continuous detection of physiological signals without the need for an external power supply is a key technology for realizing wearable electronics as next-generation biomedical devices. Herein, it is shown that a MXene/black phosphorus (BP)-based self-powered smart sensor system can be designed by integrating a flexible pressure sensor with direct-laser-writing micro-supercapacitors and solar cells. Using a layer-by-layer (LbL) self-assembly process to form a periodic interleaving MXene/BP lamellar structure results in a high energy-storage capacity in a direct-laser-writing micro-supercapacitor to drive the operation of sensors and compensate the intermittency of light illumination. Meanwhile, with MXene/BP as the sensitive layer in a flexible pressure sensor, the pressure sensitivity of the device can be improved to 77.61 kPa-1 at an optimized elastic modulus of 0.45 MPa. Furthermore, the smart sensor system with fast response time (10.9 ms) shows a real-time detection capability for the state of the human heart under physiological conditions. It is believed that the proposed study based on the design and integration of MXene materials will provide a general platform for next-generation self-powered electronics.

Hydrophobic and Stable MXene-Polymer Pressure Sensors for Wearable Electronics.

Ti3C2Tx MXene has exhibited great potential for use in wearable devices, especially as pressure sensors, due to its lamellar structure, which changes its resistance as a function of interlayer distance. Despite the good performance of the reported pure MXene pressure sensors, their practical applications are limited by moderate flexibility, excessively high MXene conductivity, and environmental effects. To address the above challenges, we incorporated multilayer MXene particles into hydrophobic poly(vinylidene fluoride) trifluoroethylene (P(VDF-TrFE)) and prepared freestanding, flexible, and stable films via spin-coating. These films were assembled into highly sensitive piezoresistive pressure sensors, which show a fast response time of 16 ms in addition to excellent long-term stability with no obvious responsivity attenuation when the sensor is exposed to air, even after 20 weeks. Moreover, the fabricated sensors could monitor human physiological signals such as knee bending and cheek bulging and could be used for speech recognition. The mapping spatial pressure distribution function was also demonstrated by the designed 10 × 10 integrated pressure sensor array platform.