Graph Convolutional Network with Self-supervised Learning for Brain Disease Classification.

Brain functional network (BFN) analysis has become a popular method for identifying neurological diseases at their early stages and revealing sensitive biomarkers related to these diseases. Due to the fact that BFN is a graph with complex structure, graph convolutional networks (GCNs) can be naturally used in the identification of BFN, and can generally achieve an encouraging performance if given large amounts of training data. In practice, however, it is very difficult to obtain sufficient brain functional data, especially from subjects with brain disorders. As a result, GCNs usually fail to learn a reliable feature representation from limited BFNs, leading to overfitting issues. In this paper, we propose an improved GCN method to classify brain diseases by introducing a self-supervised learning (SSL) module for assisting the graph feature representation. We conduct experiments to classify subjects with mild cognitive impairment (MCI) and autism spectrum disorder (ASD) respectively from normal controls (NCs). Experimental results on two benchmark databases demonstrate that our proposed scheme tends to obtain higher classification accuracy than the baseline methods.

"Multidisciplinary fast-track" care can significantly reduce risk of mortality among hip fracture patients at least 80 years old: a single-center retrospective study.

BACKGROUND: "Multidisciplinary fast-track" (MFT) care can accelerate recovery and improve prognosis after surgery, but whether it is effective in older people after hip fracture surgery is unclear. METHODS: We retrospectively compared one-year all-cause mortality between hip fracture patients at least 80 years old at our institution who underwent hip fracture surgery between January 2014 and December 2018 and who then received MFT or conventional care. Multivariable regression was used to assess the association between MFT care and mortality after adjustment for confounders. RESULTS: The final analysis included 247 patients who received MFT care and 438 who received conventional orthopedic care. The MFT group showed significantly lower one-year mortality (8.9% vs. 14.4%, P = 0.037). Log-rank testing of Kaplan-Meier survival curves confirmed the survival advantage. However, the two groups did not differ significantly in rates of mortality during hospitalization or at 30 or 90 days after surgery. Regression analysis confirmed that MFT care was associated with lower risk of one-year mortality (hazard ratio [HR] 0.47, 95% confidence interval [CI] 0.281-0.788, P = 0.04), and the survival benefit was confirmed in subgroups of patients with anemia (HR 0.453, 95% CI 0.268-0.767, P = 0.003) and patients with American Society of Anesthesiologists grade III (HR 0.202, 95% CI 0.08-0.51, P = 0.001). CONCLUSIONS: MFT care can reduce one-year mortality among hip fracture patients at least 80 years old. This finding should be verified and extended in multi-center randomized controlled trials.

Albumin Nanoparticle-Based Drug Delivery Systems.

Nanoparticle-based systems are extensively investigated for drug delivery. Among others, with superior biocompatibility and enhanced targeting capacity, albumin appears to be a promising carrier for drug delivery. Albumin nanoparticles are highly favored in many disease therapies, as they have the proper chemical groups for modification, cell-binding sites for cell adhesion, and affinity to protein drugs for nanocomplex generation. Herein, this review summarizes the recent fabrication techniques, modification strategies, and application of albumin nanoparticles. We first discuss various albumin nanoparticle fabrication methods, from both pros and cons. Then, we provide a comprehensive introduction to the modification section, including organic albumin nanoparticles, metal albumin nanoparticles, inorganic albumin nanoparticles, and albumin nanoparticle-based hybrids. We finally bring further perspectives on albumin nanoparticles used for various critical diseases.

Albumin appears to be a promising carrier for drug delivery with superior biocompatibility and enhanced targeting capacity. This review focuses on the importance of albumin nanoparticles in drug delivery and concludes the recent fabrication techniques to prepare albumin nanoparticles, the modification strategies to require functional albumin nanoparticles, and critical applications of albumin nanoparticles in various diseases. The aim of this review is to help readers understand the significant potential of albumin nanoparticles in drug delivery.

Theoretical and kinetic study of the H-atom abstraction reactions by H atom from alkyl cyclohexanes.

Reaction kinetics of hydrogen atom abstraction from six alkyl cyclohexanes, methyl cyclohexane (MCH), ethyl cyclohexane (ECH), n-propyl cyclohexane (nPCH), iso-propyl cyclohexane (iPCH), sec-butyl cyclohexane (sBCH) and iso-butyl cyclohexane (iBCH), by the H atom are systematically studied in this work. The M06-2X method combined with the 6-311++G(d,p) basis set is used to perform geometry optimization, frequency analysis and zero-point energy calculations for all species. The intrinsic reaction coordinate (IRC) calculations are performed to confirm the transition states connecting the reactants and products correctly. One-dimensional hindered rotors are used to treat the low frequency torsional models with potentials scanned at the M06-2X/6-31G level of theory. Electronic single-point energy calculations for all reactants, transition states, and products are performed at the QCISD(T)/CBS level of theory. High-pressure limiting rate constants of 39 reaction channels are obtained using conventional transition state theory with asymmetric Eckart tunneling corrections in the temperature range 298.15-2000 K. Reaction rate rules for H-atom abstraction by the H atom from alkyl cyclohexanes on primary, secondary and tertiary carbon sites on both the side chain and ring are provided. The obtained rate constants are given by the Arrhenius expression in the temperature range 500-2000 K, which can be used for the combustion kinetics model development for alkyl cyclohexanes.

LncRNA taurine upregulated gene 1 in liver disease.

Long non-coding RNAs (lncRNAs) are RNA sequences exceeding 200 nucleotides in length that lack protein-coding capacity and participate in diverse biological processes in the human body, particularly exerting a pivotal role in disease surveillance, diagnosis, and progression. Taurine upregulated gene 1 (TUG1) is a versatile lncRNA, and recent studies have revealed that the aberrant expression or function of TUG1 is intricately linked to the pathogenesis of liver diseases. Consequently, we have summarized the current understanding of the mechanism of TUG1 in liver diseases such as liver fibrosis, fatty liver, cirrhosis, liver injury, hepatitis, and liver cancer. Moreover, mounting evidence suggests that interventions targeting TUG1 or its downstream pathways may hold therapeutic promise for liver diseases. This review elucidates the characteristics, mechanisms, and targets of TUG1 in liver diseases, offering a theoretical basis for the prevention, diagnosis, treatment, and prognostic biomarkers of liver diseases.

Federated learning for medical image analysis: A survey.

Machine learning in medical imaging often faces a fundamental dilemma, namely, the small sample size problem. Many recent studies suggest using multi-domain data pooled from different acquisition sites/centers to improve statistical power. However, medical images from different sites cannot be easily shared to build large datasets for model training due to privacy protection reasons. As a promising solution, federated learning, which enables collaborative training of machine learning models based on data from different sites without cross-site data sharing, has attracted considerable attention recently. In this paper, we conduct a comprehensive survey of the recent development of federated learning methods in medical image analysis. We have systematically gathered research papers on federated learning and its applications in medical image analysis published between 2017 and 2023. Our search and compilation were conducted using databases from IEEE Xplore, ACM Digital Library, Science Direct, Springer Link, Web of Science, Google Scholar, and PubMed. In this survey, we first introduce the background of federated learning for dealing with privacy protection and collaborative learning issues. We then present a comprehensive review of recent advances in federated learning methods for medical image analysis. Specifically, existing methods are categorized based on three critical aspects of a federated learning system, including client end, server end, and communication techniques. In each category, we summarize the existing federated learning methods according to specific research problems in medical image analysis and also provide insights into the motivations of different approaches. In addition, we provide a review of existing benchmark medical imaging datasets and software platforms for current federated learning research. We also conduct an experimental study to empirically evaluate typical federated learning methods for medical image analysis. This survey can help to better understand the current research status, challenges, and potential research opportunities in this promising research field.

Source-free unsupervised domain adaptation: A survey.

Unsupervised domain adaptation (UDA) via deep learning has attracted appealing attention for tackling domain-shift problems caused by distribution discrepancy across different domains. Existing UDA approaches highly depend on the accessibility of source domain data, which is usually limited in practical scenarios due to privacy protection, data storage and transmission cost, and computation burden. To tackle this issue, many source-free unsupervised domain adaptation (SFUDA) methods have been proposed recently, which perform knowledge transfer from a pre-trained source model to the unlabeled target domain with source data inaccessible. A comprehensive review of these works on SFUDA is of great significance. In this paper, we provide a timely and systematic literature review of existing SFUDA approaches from a technical perspective. Specifically, we categorize current SFUDA studies into two groups, i.e., white-box SFUDA and black-box SFUDA, and further divide them into finer subcategories based on different learning strategies they use. We also investigate the challenges of methods in each subcategory, discuss the advantages/disadvantages of white-box and black-box SFUDA methods, conclude the commonly used benchmark datasets, and summarize the popular techniques for improved generalizability of models learned without using source data. We finally discuss several promising future directions in this field.

Recent Advances on Small-Molecule Inhibitors of Lipocalin-like Proteins.

Lipid transfer proteins (LTPs) are crucial players in nonvesicular lipid trafficking. LTPs sharing a lipocalin lipid transfer domain (lipocalin-like proteins) have a wide range of biological functions, such as regulating immune responses and cell proliferation, differentiation, and death as well as participating in the pathogenesis of inflammatory, metabolic, and neurological disorders and cancer. Therefore, the development of small-molecule inhibitors targeting these LTPs is important and has potential clinical applications. Herein, we summarize the structure and function of lipocalin-like proteins, mainly including retinol-binding proteins, lipocalins, and fatty acid-binding proteins and discuss the recent advances on small-molecule inhibitors for these protein families and their applications in disease treatment. The findings of our Perspective can provide guidance for the development of inhibitors of these LTPs and highlight the challenges that might be faced during the procedures.

Hybrid representation learning for cognitive diagnosis in late-life depression over 5 years with structural MRI.

Late-life depression (LLD) is a highly prevalent mood disorder occurring in older adults and is frequently accompanied by cognitive impairment (CI). Studies have shown that LLD may increase the risk of Alzheimer's disease (AD). However, the heterogeneity of presentation of geriatric depression suggests that multiple biological mechanisms may underlie it. Current biological research on LLD progression incorporates machine learning that combines neuroimaging data with clinical observations. There are few studies on incident cognitive diagnostic outcomes in LLD based on structural MRI (sMRI). In this paper, we describe the development of a hybrid representation learning (HRL) framework for predicting cognitive diagnosis over 5 years based on T1-weighted sMRI data. Specifically, we first extract prediction-oriented MRI features via a deep neural network, and then integrate them with handcrafted MRI features via a Transformer encoder for cognitive diagnosis prediction. Two tasks are investigated in this work, including (1) identifying cognitively normal subjects with LLD and never-depressed older healthy subjects, and (2) identifying LLD subjects who developed CI (or even AD) and those who stayed cognitively normal over five years. We validate the proposed HRL on 294 subjects with T1-weighted MRIs from two clinically harmonized studies. Experimental results suggest that the HRL outperforms several classical machine learning and state-of-the-art deep learning methods in LLD identification and prediction tasks.

Prevalence and factors associated with overweight, obesity and central obesity among adults in Shenmu City, Shaanxi Province, China.

Objectives: This research aimed to explore the prevalence and determinants of overweight, obesity, and central obesity in Shenmu City, Shaanxi Province, China and to offer guidance for preventative health measures. Methods: We conducted a multi-stage, stratified random sampling survey among 4,565 residents of Shenmu City. Data collection included questionnaires and anthropometric assessments to gather socio-demographic data and to identify cases of overweight, obesity, and central obesity. Multivariable logistic regression analysis was utilized to assess the association between various factors and these conditions. Results: The observed prevalence rates for overweight, obesity, central obesity, and the combination of overweight/obesity with central obesity were 39.9%, 18.2%, 48.0%, 32.8%, and 22.8%, respectively. Notably, the incidence of these conditions was significantly higher in men compared to women. The prevalence of overweight and obesity initially increased and then decreased with age, whereas the prevalence of central obesity consistently rose. Furthermore, a higher educational level correlated with lower prevalence rates. Additionally, our analysis indicated that hypertension, dyslipidemia, and hyperuricemia are risk factors for these conditions. Conclusions: The findings of this study offer crucial insights for formulating effective strategies to prevent and manage obesity in Shenmu City.

The prevalence of obesity and relationship between obesity indicators and chronic diseases in Northern Shaanxi, China.

INTRODUCTION: Obesity not only affects human health but also is an important risk factor for a variety of chronic diseases. Therefore, it is particularly important to analyse the epidemic trend of obesity and actively carry out the prevention and control of obesity in the population. MATERIAL AND METHODS: A total of 4565 adults were selected by multi-stage stratified random sampling in Shenmu, Shaanxi Province, China. Univariate analysis was used to explore the epidemic characteristics of obesity in this region. Multivariate logistic regression was used to analyse the relationship between obesity and chronic diseases. Finally, the prediction efficiency of different obesity indexes was analysed by drawing receiver operator characteristic curves (ROC). All statistical analysis was completed by SPSS 26.0 software. RESULTS: The prevalence rates of overweight, obesity, and central obesity were 39.9%, 18.2%, and 48.0%, respectively. After adjusting for other confounding factors, multivariate logistic regression analysis showed that overweight and obesity were risk factors for hypertension, dyslipidaemia, and hyperuricaemia. Central obesity is a risk factor for dyslipidaemia and hyperuricaemia. High level of waist-to-height ratio (WHtR) was a risk factor for dyslipidaemia and hyperuricaemia (p < 0.05). Obesity-related indicators: body mass index (BMI), waist circumference (WC), and WHtR, are strongly correlated with the increased risk of chronic diseases in northern Shaanxi, China. The optimal BMI cut-off values for predicting hypertension, dyslipidaemia, and hyperuricaemia were 24.27, 24.04, and 25.54, respectively. The optimal WC cut-off values for predicting dyslipidaemia and hyperuricaemia were 84.5 and 90.5, and WHtR cut-off values were 0.52 and 0.54, respectively. CONCLUSION: The problem of overweight, obesity, and central obesity in adults is serious in northern Shaanxi, China. Obesity of all types will increase the risk of chronic diseases. Therefore, a variety of preventive and therapeutic measures should be adopted to curb obesity and reduce the incidence of related chronic diseases.

Leveraging Brain Modularity Prior for Interpretable Representation Learning of fMRI.

Resting-state functional magnetic resonance imaging (rs-fMRI) can reflect spontaneous neural activities in the brain and is widely used for brain disorder analysis. Previous studies focus on extracting fMRI representations using machine/deep learning methods, but these features typically lack biological interpretability. The human brain exhibits a remarkable modular structure in spontaneous brain functional networks, with each module comprised of functionally interconnected brain regions-of-interest (ROIs). However, existing learning-based methods cannot adequately utilize such brain modularity prior. In this paper, we propose a brain modularity-constrained dynamic representation learning framework for interpretable fMRI analysis, consisting of dynamic graph construction, dynamic graph learning via a novel modularity-constrained graph neural network (MGNN), and prediction and biomarker detection. The designed MGNN is constrained by three core neurocognitive modules (i.e., salience network, central executive network, and default mode network), encouraging ROIs within the same module to share similar representations. To further enhance discriminative ability of learned features, we encourage the MGNN to preserve network topology of input graphs via a graph topology reconstruction constraint. Experimental results on 534 subjects with rs-fMRI scans from two datasets validate the effectiveness of the proposed method. The identified discriminative brain ROIs and functional connectivities can be regarded as potential fMRI biomarkers to aid in clinical diagnosis.

Triplet-constrained deep hashing for chest X-ray image retrieval in COVID-19 assessment.

Radiology images of the chest, such as computer tomography scans and X-rays, have been prominently used in computer-aided COVID-19 analysis. Learning-based radiology image retrieval has attracted increasing attention recently, which generally involves image feature extraction and finding matches in extensive image databases based on query images. Many deep hashing methods have been developed for chest radiology image search due to the high efficiency of retrieval using hash codes. However, they often overlook the complex triple associations between images; that is, images belonging to the same category tend to share similar characteristics and vice versa. To this end, we develop a triplet-constrained deep hashing (TCDH) framework for chest radiology image retrieval to facilitate automated analysis of COVID-19. The TCDH consists of two phases, including (a) feature extraction and (b) image retrieval. For feature extraction, we have introduced a triplet constraint and an image reconstruction task to enhance discriminative ability of learned features, and these features are then converted into binary hash codes to capture semantic information. Specifically, the triplet constraint is designed to pull closer samples within the same category and push apart samples from different categories. Additionally, an auxiliary image reconstruction task is employed during feature extraction to help effectively capture anatomical structures of images. For image retrieval, we utilize learned hash codes to conduct searches for medical images. Extensive experiments on 30,386 chest X-ray images demonstrate the superiority of the proposed method over several state-of-the-art approaches in automated image search. The code is now available online.

Vascular injury activates the ELK1/SND1/SRF pathway to promote vascular smooth muscle cell proliferative phenotype and neointimal hyperplasia.

BACKGROUND: Vascular smooth muscle cell (VSMC) proliferation is the leading cause of vascular stenosis or restenosis. Therefore, investigating the molecular mechanisms and pivotal regulators of the proliferative VSMC phenotype is imperative for precisely preventing neointimal hyperplasia in vascular disease. METHODS: Wire-induced vascular injury and aortic culture models were used to detect the expression of staphylococcal nuclease domain-containing protein 1 (SND1). SMC-specific Snd1 knockout mice were used to assess the potential roles of SND1 after vascular injury. Primary VSMCs were cultured to evaluate SND1 function on VSMC phenotype switching, as well as to investigate the mechanism by which SND1 regulates the VSMC proliferative phenotype. RESULTS: Phenotype-switched proliferative VSMCs exhibited higher SND1 protein expression compared to the differentiated VSMCs. This result was replicated in primary VSMCs treated with platelet-derived growth factor (PDGF). In the injury model, specific knockout of Snd1 in mouse VSMCs reduced neointimal hyperplasia. We then revealed that ETS transcription factor ELK1 (ELK1) exhibited upregulation and activation in proliferative VSMCs, and acted as a novel transcription factor to induce the gene transcriptional activation of Snd1. Subsequently, the upregulated SND1 is associated with serum response factor (SRF) by competing with myocardin (MYOCD). As a co-activator of SRF, SND1 recruited the lysine acetyltransferase 2B (KAT2B) to the promoter regions leading to the histone acetylation, consequently promoted SRF to recognize the specific CArG motif, and enhanced the proliferation- and migration-related gene transcriptional activation. CONCLUSIONS: The present study identifies ELK1/SND1/SRF as a novel pathway in promoting the proliferative VSMC phenotype and neointimal hyperplasia in vascular injury, predisposing the vessels to pathological remodeling. This provides a potential therapeutic target for vascular stenosis.

Preserving specificity in federated graph learning for fMRI-based neurological disorder identification.

Resting-state functional magnetic resonance imaging (rs-fMRI) offers a non-invasive approach to examining abnormal brain connectivity associated with brain disorders. Graph neural network (GNN) gains popularity in fMRI representation learning and brain disorder analysis with powerful graph representation capabilities. Training a general GNN often necessitates a large-scale dataset from multiple imaging centers/sites, but centralizing multi-site data generally faces inherent challenges related to data privacy, security, and storage burden. Federated Learning (FL) enables collaborative model training without centralized multi-site fMRI data. Unfortunately, previous FL approaches for fMRI analysis often ignore site-specificity, including demographic factors such as age, gender, and education level. To this end, we propose a specificity-aware federated graph learning (SFGL) framework for rs-fMRI analysis and automated brain disorder identification, with a server and multiple clients/sites for federated model aggregation and prediction. At each client, our model consists of a shared and a personalized branch, where parameters of the shared branch are sent to the server while those of the personalized branch remain local. This can facilitate knowledge sharing among sites and also helps preserve site specificity. In the shared branch, we employ a spatio-temporal attention graph isomorphism network to learn dynamic fMRI representations. In the personalized branch, we integrate vectorized demographic information (i.e., age, gender, and education years) and functional connectivity networks to preserve site-specific characteristics. Representations generated by the two branches are then fused for classification. Experimental results on two fMRI datasets with a total of 1218 subjects suggest that SFGL outperforms several state-of-the-art approaches.

Recent Progress in Mass Spectrometry-Based Metabolomics in Major Depressive Disorder Research.

Major depressive disorder (MDD) is a serious mental illness with a heavy social burden, but its underlying molecular mechanisms remain unclear. Mass spectrometry (MS)-based metabolomics is providing new insights into the heterogeneous pathophysiology, diagnosis, treatment, and prognosis of MDD by revealing multi-parametric biomarker signatures at the metabolite level. In this comprehensive review, recent developments of MS-based metabolomics in MDD research are summarized from the perspective of analytical platforms (liquid chromatography-MS, gas chromatography-MS, supercritical fluid chromatography-MS, etc.), strategies (untargeted, targeted, and pseudotargeted metabolomics), key metabolite changes (monoamine neurotransmitters, amino acids, lipids, etc.), and antidepressant treatments (both western and traditional Chinese medicines). Depression sub-phenotypes, comorbid depression, and multi-omics approaches are also highlighted to stimulate further advances in MS-based metabolomics in the field of MDD research.

Risk Factors for Hypocoagulability After Cardiac Surgery: A Retrospective Study.

Hemostatic disturbances after cardiac surgery can lead to excessive postoperative bleeding. Thromboelastography (TEG) was employed to evaluate perioperative coagulative alterations in patients undergoing cardiac surgery with cardiopulmonary bypass (CPB), investigating the correlation between factors concomitant with cardiac surgery and modifications in coagulation. Coagulation index as determined by TEG correlated significantly with postoperative bleeding at 24-72 h after cardiac surgery (P < .001). Among patients with a normal preoperative coagulation index, those with postoperative hypocoagulability showed significantly lower nadir temperature (P = .003), larger infused fluid volume (P = .003), and longer CPB duration (P = .033) than those with normal coagulation index. Multivariate logistic regression showed that nadir intraoperative temperature was an independent predictor of postoperative hypocoagulability (adjusted OR: 0.772, 95% CI: 0.624-0.954, P = .017). Multivariate linear regression demonstrated linear associations of nadir intraoperative temperature (P = .017) and infused fluid volume (P = .005) with change in coagulation index as a result of cardiac surgery. Patients are susceptible to hypocoagulability after cardiac surgery, which can lead to increased postoperative bleeding. Ensuring appropriate temperature and fluid volume during cardiac surgery involving CPB may reduce risk of postoperative hypocoagulability and bleeding.

Nadir Hemoglobin Concentration After Spinal Tumor Surgery: Association With Risk of Composite Adverse Events.

STUDY DESIGN: Retrospective case-control study. OBJECTIVE: To explore the association of early postoperative nadir hemoglobin with risk of a composite outcome of anemia-related and other adverse events. METHODS: We retrospectively analyzed data from spinal tumor patients who received intraoperative blood transfusion between September 1, 2013 and December 31, 2020. Uni- and multivariate logistic regression was used to explore relationships of clinicodemographic and surgical factors with risk of composite in-hospital adverse events, including death. Subgroup analysis explored the relationship between early postoperative nadir hemoglobin and composite adverse events. RESULTS: Among the 345 patients, 331 (95.9%) experienced early postoperative anemia and 69 (20%) experienced postoperative composite adverse events. Multivariate logistic regression analysis showed that postoperative nadir Hb (OR = .818, 95% CI: .672-.995, P = .044), ASA ≥3 (OR = 2.007, 95% CI: 1.086-3.707, P = .026), intraoperative RBC infusion volume (OR = 1.133, 95% CI: 1.009-1.272, P = .035), abnormal hypertension (OR = 2.199, 95% CI: 1.085-4.457, P = .029) were correlated with composite adverse events. The lumbar spinal tumor was associated with composite adverse events with a decreased odds compared to thoracic spinal tumors (OR = .444, 95% CI: .226-.876, P = .019). Compared to patients with postoperative nadir hemoglobin ≥11.0 g/dL, those with nadir <9.0 g/dL were at significantly higher risk of postoperative composite adverse events (OR = 2.709, 95% CI: 1.087-6.754, P = .032). CONCLUSION: Nadir hemoglobin <9.0 g/dL after spinal tumor surgery is associated with greater risk of postoperative composite adverse events in patients who receive intraoperative blood transfusion.

Unsupervised contrastive graph learning for resting-state functional MRI analysis and brain disorder detection.

Resting-state functional magnetic resonance imaging (rs-fMRI) helps characterize regional interactions that occur in the human brain at a resting state. Existing research often attempts to explore fMRI biomarkers that best predict brain disease progression using machine/deep learning techniques. Previous fMRI studies have shown that learning-based methods usually require a large amount of labeled training data, limiting their utility in clinical practice where annotating data is often time-consuming and labor-intensive. To this end, we propose an unsupervised contrastive graph learning (UCGL) framework for fMRI-based brain disease analysis, in which a pretext model is designed to generate informative fMRI representations using unlabeled training data, followed by model fine-tuning to perform downstream disease identification tasks. Specifically, in the pretext model, we first design a bi-level fMRI augmentation strategy to increase the sample size by augmenting blood-oxygen-level-dependent (BOLD) signals, and then employ two parallel graph convolutional networks for fMRI feature extraction in an unsupervised contrastive learning manner. This pretext model can be optimized on large-scale fMRI datasets, without requiring labeled training data. This model is further fine-tuned on to-be-analyzed fMRI data for downstream disease detection in a task-oriented learning manner. We evaluate the proposed method on three rs-fMRI datasets for cross-site and cross-dataset learning tasks. Experimental results suggest that the UCGL outperforms several state-of-the-art approaches in automated diagnosis of three brain diseases (i.e., major depressive disorder, autism spectrum disorder, and Alzheimer's disease) with rs-fMRI data.

Multi-Scale Dynamic Graph Learning for Brain Disorder Detection With Functional MRI.

Resting-state functional magnetic resonance imaging (rs-fMRI) has been widely used in the detection of brain disorders such as autism spectrum disorder based on various machine/deep learning techniques. Learning-based methods typically rely on functional connectivity networks (FCNs) derived from blood-oxygen-level-dependent time series of rs-fMRI data to capture interactions between brain regions-of-interest (ROIs). Graph neural networks have been recently used to extract fMRI features from graph-structured FCNs, but cannot effectively characterize spatiotemporal dynamics of FCNs, e.g., the functional connectivity of brain ROIs is dynamically changing in a short period of time. Also, many studies usually focus on single-scale topology of FCN, thereby ignoring the potential complementary topological information of FCN at different spatial resolutions. To this end, in this paper, we propose a multi-scale dynamic graph learning (MDGL) framework to capture multi-scale spatiotemporal dynamic representations of rs-fMRI data for automated brain disorder diagnosis. The MDGL framework consists of three major components: 1) multi-scale dynamic FCN construction using multiple brain atlases to model multi-scale topological information, 2) multi-scale dynamic graph representation learning to capture spatiotemporal information conveyed in fMRI data, and 3) multi-scale feature fusion and classification. Experimental results on two datasets show that MDGL outperforms several state-of-the-art methods.