Functional brain network identification and fMRI augmentation using a VAE-GAN framework.

Recently, deep learning models have achieved superior performance for mapping functional brain networks from functional magnetic resonance imaging (fMRI) data compared with traditional methods. However, due to the lack of sufficient data and the high dimensionality of brain volume, deep learning models of fMRI tend to suffer from overfitting. In addition, existing methods rarely studied fMRI data augmentation and its application. To address these issues, we developed a VAE-GAN framework that combined a VAE (variational auto-encoder) with a GAN (generative adversarial net) for functional brain network identification and fMRI augmentation. As a generative model, the VAE-GAN models the distribution of fMRI so that it enables the extraction of more generalized features, and thus relieve the overfitting issue. The VAE-GAN is easier to train on fMRI than a standard GAN since it uses latent variables from VAE to generate fake data rather than relying on random noise that is used in a GAN, and it can generate higher quality of fake data than VAE since the discriminator can promote the training of the generator. In other words, the VAE-GAN inherits the advantages of VAE and GAN and avoids their limitations in modeling of fMRI data. Extensive experiments on task fMRI datasets from HCP have proved the effectiveness and superiority of the proposed VAE-GAN framework for identifying both temporal features and functional brain networks compared with existing models, and the quality of fake data is higher than those from VAE and GAN. The results on resting state fMRI of Attention Deficit Hyperactivity Disorder (ADHD)-200 dataset further demonstrated that the fake data generated by the VAE-GAN can help improve the performance of brain network modeling and ADHD classification.

A deep learning method for autism spectrum disorder identification based on interactions of hierarchical brain networks.

BACKGROUND: It has been recently shown that deep learning models exhibited remarkable performance of representing functional Magnetic Resonance Imaging (fMRI) data for the understanding of brain functional activities. With hierarchical structure, deep learning models can infer hierarchical functional brain networks (FBN) from fMRI. However, the applications of the hierarchical FBNs have been rarely studied. METHODS: In this work, we proposed a hierarchical recurrent variational auto-encoder (HRVAE) to unsupervisedly model the fMRI data. The trained HRVAE encoder can predict hierarchical temporal features from its three hidden layers, and thus can be regarded as a hierarchical feature extractor. Then LASSO (least absolute shrinkage and selection operator) regression was applied to estimate the corresponding hierarchical FBNs. Based on the hierarchical FBNs from each subject, we constructed a novel classification framework for brain disorder identification and test it on the Autism Brain Imaging Data Exchange (ABIDE) dataset, a world-wide multi-site database of autism spectrum disorder (ASD). We analyzed the hierarchy organization of FBNs, and finally used the overlaps of hierarchical FBNs as features to differentiate ASD from typically developing controls (TDC). RESULTS: The experimental results on 871 subjects from ABIDE dataset showed that the HRVAE model can effectively derive hierarchical FBNs including many well-known resting state networks (RSN). Moreover, the classification result improved the state-of-the-art by achieving a very high accuracy of 82.1 %. CONCLUSIONS: This work presents a novel data-driven deep learning method using fMRI data for ASD identification, which could provide valuable reference for clinical diagnosis. The classification results suggest that the interactions of hierarchical FBNs have association with brain disorder, which promotes the understanding of FBN hierarchy and could be applied to other brain disorder analysis.

Isolation and identification of the alga-symbiotic bacterium Gordonia and characterisation of its exopolysaccharide.

An exopolysaccharide (EPS)-producing bacterium TD18, isolated from the culture broth of green alga Scenedesmus obliquus, was identified as Gordonia terrae based on the 100% identity of 16S rRNA sequences and designated Gordonia terrae TD18. The results of compositional and structural analyses and physiochemical tests show that (1) the exopolysaccharide produced by G. terrae TD18 (TD18-EPS) is an acidic hetero-polysaccharide with a molecular weight of 23 kDa, consisting of glucose, mannose, galactose and glucuronic acid, and (2) TD18-EPS is of high thermal stability with a degradation temperature of 308 °C, the solution of which is non-Newtonian pseudoplastic fluid exhibiting good emulsifying properties over a wide range of temperatures, pH and NaCl concentrations. Hence, Gordonia terrae TD18 is the first alga-symbiotic Gordonia strain identified thus far, while TD18-EPS is unique in terms of composition and structure, different from the known Gordonia EPS, with excellent physiochemical properties and thus has potential applications in industry.

A comparative study of thumb reconstruction through the transplant of the first toe compound free flap between emergency surgery and elective surgery.

This study compared emergency surgery with elective surgery for thumb reconstruction to explore the advantages, safety, and clinical value of emergency reconstruction. By comparing the advantages and disadvantages of thumb reconstruction in emergency surgery and elective surgery, it provides data support for optimizing the treatment process and methods. In this study, 22 patients who underwent thumb reconstruction in Rizhao people's Hospital from January 2018 to December 2020 were randomly divided into emergency operation group and elective operation group. The differences in operation period, hospitalization time, postoperative complications, hand function score, and satisfaction score between the 2 groups were analyzed. The operation period and hospitalization time of patients in the emergency surgery group were significantly lower than those in the elective surgery group, with statistical significance (P < .05). There was no significant difference in postoperative complications between the 2 groups (P > .05). After 3 months of rehabilitation training, the 2-point discrimination, functional score, and satisfaction score of the reconstructed thumb in the emergency surgery group were higher than those in the elective surgery group, and the difference was statistically significant (P < .05). Emergency reconstruction of the thumb can reduce operation time and hospitalization time, reduce operation costs, and obtain a more ideal appearance and function.

Analysis of clinical characteristics, pathogen infection, and drug sensitivity of Marine injury patients: A cross-sectional study.

The infection rate is high in patients injured at sea, and because of the unique distribution of marine microorganisms, the infection is often not easily controlled effectively with the empirical application of antibiotics. This study aims to consider the clinical characteristics and pathogen infection and drug susceptibility of patients injured at sea. From 2019 to 2021, there were 635 patients injured at sea in Rizhao People's Hospital. We assess the patient's basic condition, while performing bacterial culture and drug susceptibility testing on wound exudate or pus from infected patients. Among the 635 patients injured at sea, 195 people were infected, and the infection rate was 30.71%. Infected patients are usually older, have longer prehospital visits, and have lower normal levels of red blood cells, hemoglobin, total protein, and albumin. The causes of injury in infected patients were mainly avulsion and puncture injuries, and the types of injuries were mainly bone fracture, vascular injury, and nerve injury. A total of 305 strains of pathogenic bacteria were cultured in 195 patients. Gram-negative bacteria accounted for 77.05% (235 strains), of which Proteus was the most. Gram-positive bacteria accounted for 22.95% (70 strains), of which Staphylococcus aureus was the most. Gram-negative bacilli were sensitive to aminoglycosides, lactam antibiotics, carbapenems antibiotics, sulfonamides, quinolones, fourth-generation cephalosporins, and antibacterial drugs containing enzyme inhibitors, while most of the bacteria were resistant to penicillins, first-generation cephalosporins, and second-generation cephalosporins. Gram-positive bacteria were sensitive to quinuptin/dafoptin, rifampicin, linezolid, gentamicin, tigacycline, and vancomycin but resistant to penicillin antibiotics. Due to the particularity of marine injuries, patients are prone to infection. Pathogen culture and drug sensitivity analysis play an important role in guiding antiinfective treatment for marine injured patients.

Learning brain representation using recurrent Wasserstein generative adversarial net.

BACKGROUND AND OBJECTIVE: To understand brain cognition and disorders, modeling the mapping between mind and brain has been of great interest to the neuroscience community. The key is the brain representation, including functional brain networks (FBN) and their corresponding temporal features. Recently, it has been proven that deep learning models have superb representation power on functional magnetic resonance imaging (fMRI) over traditional machine learning methods. However, due to the lack of high-quality data and labels, deep learning models tend to suffer from overfitting in the training process. METHODS: In this work, we applied a recurrent Wasserstein generative adversarial net (RWGAN) to learn brain representation from volumetric fMRI data. Generative adversarial net (GAN) is widely used in natural image generation and is able to capture the distribution of the input data, which enables the extraction of generalized features from fMRI and thus relieves the overfitting issue. The recurrent layers in RWGAN are designed to better model the local temporal features of the fMRI time series. The discriminator of RWGAN works as a deep feature extractor. With LASSO regression, the RWGAN model can decompose the fMRI data into temporal features and spatial features (FBNs). Furthermore, the generator of RWGAN can generate high-quality new data for fMRI augmentation. RESULTS: The experimental results on seven tasks from the HCP dataset showed that the RWGAN can learn meaningful and interpretable temporal features and FBNs, compared to HCP task designs and general linear model (GLM) derived networks. Besides, the results on different training datasets showed that the RWGAN performed better on small datasets than other deep learning models. Moreover, we used the generator of RWGAN to yield fake subjects. The result showed that the fake data can also be used to learn meaningful representation compared to those learned from real data. CONCLUSIONS: To our best knowledge, this work is among the earliest attempts of applying generative deep learning for modeling fMRI data. The proposed RWGAN offers a novel methodology for learning brain representation from fMRI, and it can generate high-quality fake data for the potential use of fMRI data augmentation.

Separation, quantification and characterisation of the pigment produced by Paecilomyces lilacinus TD16.

Fungal pigments are important natural products with a wide range of applications. In this study, the purple-red pigment produced by the fungus Paecilomyces lilacinus TD16 (TD16 pigment) was separated with acidulated ethyl acetate and purified by silica gel column chromatography. Results of UV-visible spectrum and HPLC analyses showed that TD16 pigment is a new polyketide pigment with three absorption peaks at 228, 272 and 527 nm and a retention time of 11.4665 min distinct from those of other Paecilomyces-sourced pigments. Results of kinetic analysis and antimicrobial activity assay showed that TD16 pigment is a non-growth-associated secondary products with broad-spectrum antimicrobial activity on both bacteria and fungi and thus of potential application in industry.[Formula: see text].

NEAT1 silencing alleviates pulmonary arterial smooth muscle cell migration and proliferation under hypoxia through regulation of miR­34a­5p/KLF4 in vitro.

Pulmonary arterial hypertension (PAH) is a severe vascular disease that adversely affects patient health and can be life threatening. The present study aimed to investigate the detailed role of nuclear paraspeckle assembly transcript 1 (NEAT1) in PAH. Using RT­qPCR, the expression levels of NEAT1, microRNA (miR)­34a­5p, and Krüppel­like factor 4 (KLF4) were detected in both hypoxia­treated pulmonary arterial smooth muscle cells (PASMCs) and serum from PAH patients. Then, the interactions among miR­34a­5p, NEAT1, and KLF4 were evaluated by dual­luciferase reporter assay. The detailed role of the NEAT1/miR­34a­5p/KLF4 axis in PAH pathogenesis was further explored using MTT, Transwell, and western blot assays. The results revealed that NEAT1 targeted miR­34a­5p and miR­34a­5p targeted KLF4. In hypoxia­treated PASMCs and serum from PAH patients, high NEAT1 and KLF4 expression levels and low miR­34a­5p expression were observed. The proliferation and migration of hypoxia­treated PASMCs were reduced by transfection with sh­NEAT1 or miR­34a­5p mimics. The suppressive effects of NEAT1 knockdown on the proliferation and migration of hypoxia­treated PASMCs were reversed by knock down of miR­34a­5p expression and increased KLF4 expression. NEAT1 was not only highly expressed in the serum of PAH patients but its silencing also alleviated PAH by regulating miR­34a­5p/KLF4 in vitro. The present study highlighted a potential new therapeutic target and diagnostic biomarker for PAH.

Modeling and augmenting of fMRI data using deep recurrent variational auto-encoder.

Objective. Recently, deep learning models have been successfully applied in functional magnetic resonance imaging (fMRI) modeling and associated applications. However, there still exist at least two challenges. Firstly, due to the lack of sufficient data, deep learning models tend to suffer from overfitting in the training process. Secondly, it is still challenging to model the temporal dynamics from fMRI, due to that the brain state is continuously changing over scan time. In addition, existing methods rarely studied and applied fMRI data augmentation.Approach. In this work, we construct a deep recurrent variational auto-encoder (DRVAE) that combined variational auto-encoder and recurrent neural network, aiming to address all of the above mentioned challenges. The encoder of DRVAE can extract more generalized temporal features from assumed Gaussian distribution of input data, and the decoder of DRVAE can generate new data to increase training samples and thus partially relieve the overfitting issue. The recurrent layers in DRVAE are designed to effectively model the temporal dynamics of functional brain activities. LASSO (least absolute shrinkage and selection operator) regression is applied on the temporal features and input fMRI data to estimate the corresponding spatial networks.Main results. Extensive experimental results on seven tasks from HCP dataset showed that the DRVAE and LASSO framework can learn meaningful temporal patterns and spatial networks from both real data and generated data. The results on group-wise data and single subject suggest that the brain activities may follow certain distribution. Moreover, we applied DRVAE on four resting state fMRI datasets from ADHD-200 for data augmentation, and the results showed that the classification performances on augmented datasets have been considerably improved.Significance. The proposed method can not only derive meaningful temporal features and spatial networks from fMRI, but also generate high-quality new data for fMRI data augmentation and associated applications.

Simultaneous spatial-temporal decomposition for connectome-scale brain networks by deep sparse recurrent auto-encoder.

Exploring the spatial patterns and temporal dynamics of human brain activity has been of great interest, in the quest to better understand connectome-scale brain networks. Though modeling spatial and temporal patterns of functional brain networks have been researched for a long time, the development of a unified and simultaneous spatial-temporal model has yet to be realized. For instance, although some deep learning methods have been proposed recently in order to model functional brain networks, most of them can only represent either spatial or temporal perspective of functional Magnetic Resonance Imaging (fMRI) data and rarely model both domains simultaneously. Due to the recent success in applying sequential auto-encoders for brain decoding, in this paper, we propose a deep sparse recurrent auto-encoder (DSRAE) to be applied unsupervised to learn spatial patterns and temporal fluctuations of brain networks at the same time. The proposed DSRAE was evaluated and validated based on three tasks of the publicly available Human Connectome Project (HCP) fMRI dataset, resulting with promising evidence. To the best of our knowledge, the proposed DSRAE is among the early efforts in developing unified models that can extract connectome-scale spatial-temporal networks from 4D fMRI data simultaneously.

Long-read transcriptomic analysis of orb-weaving spider Araneus ventricosus indicates transcriptional diversity of spidroins.

Spider silk, which is composed of diverse silk proteins (spidroin), is a kind of natural high-mass biomaterial with great potential. However, due to the complexity of both the structure and the composition of the spidroins in natural spider silk, application of this valuable biomass is still limited to date. There are diverse kinds of spider silk in the orb-weaving spider with different mechanical and structural characteristics. In order to systematically illustrate the landscape of all the different spidrons, here we chose Araneus ventricosus, an orb-weaving spider with superior silk mechanical features and genome information, to generate a long-read whole body transcriptome. We deciphered the repeat arrangements of each kind of spidroin, based on which we found that there are substantially transcriptional diversity of each spidroin gene. Some repeat motifs are not documented before. Specifically, we discovered novel full-lengh MaSp transcript as well as a relatively small full-length AcSp isoforms, which are potential promising materials for bioengineering of recombinant spidroin. Our study provided a batch of new spidron resources with detail sequential information. The finding of transcriptional diversity may provide cues in understanding of within-species variation of the mechanical properties of the natural spider silk and further molecular designing of recombinant spidroin.

NR4A2 Exacerbates Cerebral Ischemic Brain Injury via Modulating microRNA-652/Mul1 Pathway.

BACKGROUND: Nuclear receptor subfamily group A member 2 (NR4A2), a transcription factor, was suggested to be involved in the pathogenesis of ischemic stroke. Nevertheless, the specific role of NR4A2 in ischemic brain injury has yet to be elucidated. Our aim was to probe the mechanisms behind the repression of microRNA (miRNA) expression resulting from NR4A2 regulation in ischemic brain injury. METHODS: A rat model with transient global cerebral ischemia (tGCI) was established, followed by HE staining and immunohistochemistry for verification. Subsequently, NR4A2 expression in rat brain tissues was detected by RT-qPCR, Western blot and immunohistochemistry. Then, PC12 cells were treated with NR4A2 alteration and subjected to oxygen-glucose deprivation (OGD) for cerebral ischemia simulation. Cell viability, apoptosis and cycle distribution were detected by CCK-8 and flow cytometry, respectively. miR-652 expression in rat brain tissues and cells was then detected by RT-qPCR, and then the targeting mRNAs of miR-652 were predicted through bioinformatic websites. Finally, the effect of miR-652 and mitochondrial E3 ubiquitin ligase 1 (Mul1) on the PC12 cell activity after OGD treatment was verified by rescue experiments. RESULTS: NR4A2 and Mul1 were expressed highly in brain tissues of rats with tGCI, while miR-652 was expressed poorly. NR4A2 inhibited the expression of miR-652 by transcription, thus blocking the inhibition of miR-652 on Mul1 to repress PC12 cell activity and promote apoptosis and G0/G1 cell cycle arrest. CONCLUSION: The transcription factor NR4A2 mediates the expression of Mul1 through transcriptional repression of miR-652, thus promoting ischemic brain injury.

Modeling task-based fMRI data via deep belief network with neural architecture search.

It has been shown that deep neural networks are powerful and flexible models that can be applied on fMRI data with superb representation ability over traditional methods. However, a challenge of neural network architecture design has also attracted attention: due to the high dimension of fMRI volume images, the manual process of network model design is very time-consuming and not optimal. To tackle this problem, we proposed an unsupervised neural architecture search (NAS) framework on a deep belief network (DBN) that models volumetric fMRI data, named NAS-DBN. The NAS-DBN framework is based on Particle Swarm Optimization (PSO) where the swarms of neural architectures can evolve and converge to a feasible optimal solution. The experiments showed that the proposed NAS-DBN framework can quickly find a robust architecture of DBN, yielding a hierarchy organization of functional brain networks (FBNs) and temporal responses. Compared with 3 manually designed DBNs, the proposed NAS-DBN has the lowest testing loss of 0.0197, suggesting an overall performance improvement of up to 47.9 %. For each task, the NAS-DBN identified 260 FBNs, including task-specific FBNs and resting state networks (RSN), which have high overlap rates to general linear model (GLM) derived templates and independent component analysis (ICA) derived RSN templates. The average overlap rate of NAS-DBN to GLM on 20 task-specific FBNs is as high as 0.536, indicating a performance improvement of up to 63.9 % in respect of network modeling. Besides, we showed that the NAS-DBN can simultaneously generate temporal responses that resemble the task designs very well, and it was observed that widespread overlaps between FBNs from different layers of NAS-DBN model form a hierarchical organization of FBNs. Our NAS-DBN framework contributes an effective, unsupervised NAS method for modeling volumetric task fMRI data.

Colorless Silk/Copper Sulfide Hybrid Fiber and Fabric with Spontaneous Heating Property under Sunlight.

With the increasing demand for comfort, thinness, and warmth of fabrics, various functional fibers have emerged. However, natural silkworm silk, as one of the most widely used natural fibers in textile, faces the issue that it cannot be modified during the spinning process like synthetic fibers. Herein, copper sulfide nanoparticles (CuS NPs) with a near-infrared (NIR) absorption property were first prepared by using regenerated silk fibroin (RSF) as the biological template. Then, trace CuS NPs prepared in RSF solution (no more than 100 ppm) were added into the RSF spinning dope to prepare colorless RSF/CuS hybrid fibers via wet-spinning process. The tensile test of the RSF/CuS hybrid fibers showed that the toughness was improved with the addition of CuS NPs, which completely met the requirements of textile development. The temperature of RSF/CuS hybrid fiber bundles could increase 18.5 °C within 3 min under 1064 nm laser irradiation with power density of 1.0 W/cm2. Finally, these RSF/CuS hybrid fiber bundles were woven into silk fabric or embroidered on a cotton fabric. Under the simulated sunlight, the temperature of RSF/CuS fabric could increase to more than 40 °C from room temperature. Also, as per the infrared images, the pattern of embroidery displayed a significant difference in temperature increase as compared to cotton matrix. Based on these results, an almost colorless RSF/CuS hybrid fiber that can be mass produced by wet spinning may have great potential in the fabrication of dyeable, light, and comfortable silk functional fabric with spontaneous heating characteristics under sunlight.

Artificial ligament made from silk protein/Laponite hybrid fibers.

With developments in tissue engineering, artificial ligaments are expected to be future materials for anterior cruciate ligament (ACL) reconstruction. However, poor healing of the intraosseous part after ACL reconstruction significantly hinders their applications in this field. In this study, a bioactive clay Laponite (LAP) was introduced into the regenerated silk fibroin (RSF) spinning dope to produce functional RSF/LAP hybrid fibers by wet-spinning. These RSF/LAP hybrid fibers were then woven into artificial ligament for ACL reconstruction. The structure and mechanical properties of RSF/LAP hybrid fibers were extensively studied by different means. Results confirmed the presence of LAP in RSF fibers and revealed that the addition of LAP slightly deteriorated the comprehensive mechanical properties of RSF fibers. However, they were still much tougher (with higher breaking energy) than those of degummed natural silkworm silk that was earlier used for making artificial ligament. The artificial ligament woven from RSF/LAP hybrid fibers showed better cytocompatibility and osteogenic differentiation with mouse pre-osteoblasts in vitro than those made from degummed natural silkworm silks and pure RSF fibers. Furthermore, in vivo study in a rat ACL reconstruction model demonstrated that the presence of LAP in the artificial ligament could significantly enhance the graft osseointegration process and also improve the corresponding biomechanical properties of the artificial ligament. Based upon these results, the RSF/LAP hybrid fibers, which can be mass produced by wet-spinning process, are believed to have a great potential for use as artificial ligament materials for ACL reconstruction. STATEMENT OF SIGNIFICANCE: In this study, we successfully introduced Laponite (LAP), a kind of clay that has the function of osteogenic induction, into regenerated silk fibroin (RSF) fibers, which was prepared by a mature wet-spinning method developed in our lab. We believe that through artificial spinning, additional functional components can be added into RSF fibers, which one can hardly achieve with natural silks. We showed that the artificial ligament woven from RSF/LAP hybrid fibers had better cytocompatibility and osteogenic differentiation for mouse pre-osteoblasts in vitro, and significantly enhanced the graft osseointegration process and improved the corresponding biomechanical properties in a rat ACL reconstruction model in vivo, compared to those artificial ligaments made from degummed natural silkworm silks and pure RSF fibers.

Hierarchical Organization of Functional Brain Networks Revealed by Hybrid Spatiotemporal Deep Learning.

Hierarchical organization of brain function has been an established concept in the neuroscience field for a long time, however, it has been rarely demonstrated how such hierarchical macroscale functional networks are actually organized in the human brain. In this study, to answer this question, we propose a novel methodology to provide an evidence of hierarchical organization of functional brain networks. This article introduces the hybrid spatiotemporal deep learning (HSDL), by jointly using deep belief networks (DBNs) and deep least absolute shrinkage and selection operator (LASSO) to reveal the temporal hierarchical features and spatial hierarchical maps of brain networks based on the Human Connectome Project 900 functional magnetic resonance imaging (fMRI) data sets. Briefly, the key idea of HSDL is to extract the weights between two adjacent layers of DBNs, which are then treated as the hierarchical dictionaries for deep LASSO to identify the corresponding hierarchical spatial maps. Our results demonstrate that both spatial and temporal aspects of dozens of functional networks exhibit multiscale properties that can be well characterized and interpreted based on existing computational tools and neuroscience knowledge. Our proposed novel hybrid deep model is used to provide the first insightful opportunity to reveal the potential hierarchical organization of time series and functional brain networks, using task-based fMRI signals of human brain.

Modeling Hierarchical Brain Networks via Volumetric Sparse Deep Belief Network.

It has been recently shown that deep learning models such as convolutional neural networks (CNN), deep belief networks (DBN) and recurrent neural networks (RNN), exhibited remarkable ability in modeling and representing fMRI data for the understanding of functional activities and networks because of their superior data representation capability and wide availability of effective deep learning tools. For example, spatial and/or temporal patterns of functional brain activities embedded in fMRI data can be effectively characterized and modeled by a variety of CNN/DBN/RNN deep learning models as shown in recent studies. However, it has been rarely investigated whether it is possible to directly infer hierarchical brain networks from volumetric fMRI data using deep learning models such as DBN. The perceived difficulties of such studies include very large number of input variables, very large number of training parameters, the lack of effective software tools, the challenge of results interpretation, and etc. To bridge these technical gaps, we designed a novel volumetric sparse deep belief network (VS-DBN) model and implemented it through the popular TensorFlow open source platform to reconstruct hierarchical brain networks from volumetric fMRI data based on the Human Connectome Project (HCP) 900 subjects release. Our experimental results showed that a large number of interpretable and meaningful brain networks can be robustly reconstructed from HCP 900 subjects in a hierarchical fashion, and importantly, these brain networks exhibit reasonably good consistency and correspondence across multiple HCP task-based fMRI datasets. Our work contributed a new general deep learning framework for inferring multiscale volumetric brain networks and offered novel insights into the hierarchical organization of functional brain architecture.

Effect of stress on the molecular structure and mechanical properties of supercontracted spider dragline silks.

Supercontraction is one of the most interesting properties of spider dragline silks. In this study, changes in the secondary structures of the Nephila edulis spider dragline silk after it was subjected to different supercontraction processes were investigated by integrating synchrotron Fourier transform infrared (S-FTIR) microspectroscopy and mechanical characterization. The results showed that after free supercontraction, the ß-sheet lost most of its orientation, while the helix and random coils were almost totally disordered. Interestingly, by conducting different types of supercontractions (i.e., stretching of the free supercontracted spider dragline silk to its original length or performing constrained supercontraction), it was found that although the molecular structures all changed after supercontraction, the mechanical properties almost remained unchanged when the length of the spider dragline silk did not change significantly. The other interesting conclusion obtained is that the manual stretching of a poorly oriented spider dragline silk cannot selectively improve the orientation degree of the ß-sheet in the spider silk, but increase the orientation degree of all conformations (ß-sheet, helix, and random). These experimental findings not only help to unveil the structure-property-function relationship of natural spider silks, but also provide a useful guideline for the design of biomimetic spider fiber materials.

Discovering hierarchical common brain networks via multimodal deep belief network.

Studying a common architecture reflecting both brain's structural and functional organizations across individuals and populations in a hierarchical way has been of significant interest in the brain mapping field. Recently, deep learning models exhibited ability in extracting meaningful hierarchical structures from brain imaging data, e.g., fMRI and DTI. However, deep learning models have been rarely used to explore the relation between brain structure and function yet. In this paper, we proposed a novel multimodal deep believe network (DBN) model to discover and quantitatively represent the hierarchical organizations of common and consistent brain networks from both fMRI and DTI data. A prominent characteristic of DBN is that it is capable of extracting meaningful features from complex neuroimaging data with a hierarchical manner. With our proposed DBN model, three hierarchical layers with hundreds of common and consistent brain networks across individual brains are successfully constructed through learning a large dimension of representative features from fMRI/DTI data.

Recognizing Brain States Using Deep Sparse Recurrent Neural Network.

Brain activity is a dynamic combination of different sensory responses and thus brain activity/state is continuously changing over time. However, the brain's dynamical functional states recognition at fast time-scales in task fMRI data have been rarely explored. In this paper, we propose a novel 5-layer deep sparse recurrent neural network (DSRNN) model to accurately recognize the brain states across the whole scan session. Specifically, the DSRNN model includes an input layer, one fully-connected layer, two recurrent layers, and a softmax output layer. The proposed framework has been tested on seven task fMRI data sets of Human Connectome Project. Extensive experiment results demonstrate that the proposed DSRNN model can accurately identify the brain's state in different task fMRI data sets and significantly outperforms other auto-correlation methods or non-temporal approaches in the dynamic brain state recognition accuracy. In general, the proposed DSRNN offers a new methodology for basic neuroscience and clinical research.