Dynamic brain networks provide a powerful representation for capturing temporal variations in functional brain connectivity and have gained increasing attention in brain disease diagnosis. However, most existing methods extract features from isolated time windows, making it difficult to capture the high-order dynamic evolution of brain activity. Moreover, these methods often neglect the functional heterogeneity among brain regions, thereby limiting diagnostic performance. To address these limitations, we propose HyperDiag, a novel temporal-regional Hypergraph learning via topology-enhanced state propagation for brain disease Diagnosis. Specifically, we first design a dual-level hypergraph learning strategy: a temporally-evolving hypergraph message passing strategy to capture dynamic high-order dependencies within and across time windows, and meanwhile, a region-wise functional hypergraph learning strategy to capture regional dependencies. Subsequently, we construct a topology-enhanced selective state-space propagation network to integrate complementary information from both the temporally-evolving and region-wise features. Extensive experiments on four brain disorder datasets (ABIDE-I, ADNI, REST-meta-MDD, and Epilepsy) demonstrate that HyperDiag not only outperforms state-of-the-art methods but also identifies biologically meaningful abnormal connections, offering potential biomarkers for clinical interpretation.

Our ability to localize low‐frequency sounds relies on interaural time differences (ITD), a sensitivity that is fundamental to our ability to pinpoint sound sources in azimuth and segregate competing sound sources across a wide range of situations. However, the neural representation of ITD has not been well characterized. This study aimed to elucidate distinctions in evoked cortical potentials and global functional connectivity during the unattended processing of sound localization on the horizontal plane, by recording and analyzing the auditory mismatch negativity (MMN) in adults, employing a deviant‐standard oddball paradigm. A centrally‐positioned sound source at the midline of the horizontal plane (ITD = 0 μs) served as the standard stimulus, while lateralized sounds with varying ITDs constituted the deviants. Results of MMN characteristics and the distribution of theta band power revealed a contralateral regulation mechanism of sound localization. To delve deeper into functional connectivity dynamics among different deviant stimulus groups, we computed the phase lag index within the theta band. Augmented functional connectivity was found between frontal electrode pairs when sound stimuli were directed towards the central compared to peripheral locations. In addition, assessments of global efficiency demonstrated that the peripheral sound stimuli revealed a higher global efficiency for peripheral sound stimuli. These observations suggest that smaller deviation from the center angle engages enhanced top‐down attentional modulation to salient features. In summary, our results reinforced the contralateral regulatory mechanism governing sound source localization and illuminated the unique characteristics of theta band neural responses.

IntroductionCleft lip and/or palate (CLP) patients still have severe speech disorder requiring speech rehabilitation after surgical repair. The clarity of language rehabilitation is evaluated clinically by the Language Rehabilitation Scale. However, the pattern and underlying mechanisms of functional changes in the brain are not yet clear. Recent studies suggest that the brain’s reconfiguration efficiency appears to be a key feature of its network dynamics and general cognitive abilities. In this study, we compared the association between rehabilitation effects and reconfiguration efficiency.MethodsWe evaluated CLP patients with speech rehabilitation (n = 23) and without speech rehabilitation (n = 23) and normal controls (n = 25). Assessed CLP patients on the Chinese Speech Intelligibility Test Word Lists and collected fMRI data and behavioral data for all participants. We compared behavioral data and task activation levels between participants for between-group differences and calculated reconfiguration efficiencies for each task based on each participant. In patients, we correlated reconfiguration efficiency with task performance and measured the correlation between them.ResultsBehaviorally, CLP patients with rehabilitation scored significantly higher than those without rehabilitation on the Chinese Speech Intelligibility Test Word Lists. Rehabilitation caused local brain activation levels of CLP patients to converge toward those of controls, indicating rehabilitative effects on brain function. Analysis of reconfiguration efficiency across tasks at the local and whole-brain levels identified underlying recovery mechanisms. Whole-brain reconfiguration efficiency was significantly and positively correlated with task performance.ConclusionOur results suggest that speech rehabilitation can improve the level of language-related brain activity in CLP patients, and that reconfiguration efficiency can be used as an assessment index of language clarity to evaluate the effectiveness of brain rehabilitation in CLP patients, a finding that can provide a better understanding of the degree of brain function recovery in patients.

BackgroundPatients with age-related hearing loss (ARHL) often struggle with tracking and locating sound sources, but the neural signature associated with these impairments remains unclear.Materials and methodsUsing a passive listening task with stimuli from five different horizontal directions in functional magnetic resonance imaging, we defined functional regions of interest (ROIs) of the auditory “where” pathway based on the data of previous literatures and young normal hearing listeners (n = 20). Then, we investigated associations of the demographic, cognitive, and behavioral features of sound localization with task-based activation and connectivity of the ROIs in ARHL patients (n = 22).ResultsWe found that the increased high-level region activation, such as the premotor cortex and inferior parietal lobule, was associated with increased localization accuracy and cognitive function. Moreover, increased connectivity between the left planum temporale and left superior frontal gyrus was associated with increased localization accuracy in ARHL. Increased connectivity between right primary auditory cortex and right middle temporal gyrus, right premotor cortex and left anterior cingulate cortex, and right planum temporale and left lingual gyrus in ARHL was associated with decreased localization accuracy. Among the ARHL patients, the task-dependent brain activation and connectivity of certain ROIs were associated with education, hearing loss duration, and cognitive function.ConclusionConsistent with the sensory deprivation hypothesis, in ARHL, sound source identification, which requires advanced processing in the high-level cortex, is impaired, whereas the right–left discrimination, which relies on the primary sensory cortex, is compensated with a tendency to recruit more resources concerning cognition and attention to the auditory sensory cortex. Overall, this study expanded our understanding of the neural mechanisms contributing to sound localization deficits associated with ARHL and may serve as a potential imaging biomarker for investigating and predicting anomalous sound localization.

In multi-site brain disease diagnosis studies, traditional centralized training methods necessitate sharing medical data, posing significant privacy risks. Federated learning (FL) offers a privacy-preserving solution by enabling global model training through aggregating locally trained models from multiple data centers without sharing raw data. However, current FL approaches rely on a server-based network topology, where central server failure disrupts training. Additionally, data heterogeneity across sites often slows convergence and reduces accuracy. To overcome these issues, we introduce a decentralized personalized federated learning collaborative aggregation network (pFedCAN). This framework has two core components: (1) separating local models into shared and personalized layers, and (2) forming a collaborative aggregation network via similarity detection in the shared layers. Specifically, each center trains its local model, then separates it into shared and personalized layers. The shared layer is exchanged with other centers, while the personalized layer remains local. Data centers analyze similarities in received shared layers to build a collaborative network, where shared layers from similar centers are aggregated to refine the model. This approach flexibly adapts to varying levels of data heterogeneity, enhancing model training efficiency. Validation on public datasets, ABIDE I and ADHD, shows that the proposed method outperforms current leading techniques.

Previous studies reported that auditory cortices (AC) were mostly activated by sounds coming from the contralateral hemifield. As a result, sound locations could be encoded by integrating opposite activations from both sides of AC (“opponent hemifield coding”). However, human auditory “where” pathway also includes a series of parietal and prefrontal regions. It was unknown how sound locations were represented in those high-level regions during passive listening. Here, we investigated the neural representation of sound locations in high-level regions by voxel-level tuning analysis, regions-of-interest-level (ROI-level) laterality analysis, and ROI-level multivariate pattern analysis. Functional magnetic resonance imaging data were collected while participants listened passively to sounds from various horizontal locations. We found that opponent hemifield coding of sound locations not only existed in AC, but also spanned over intraparietal sulcus, superior parietal lobule, and frontal eye field (FEF). Furthermore, multivariate pattern representation of sound locations in both hemifields could be observed in left AC, right AC, and left FEF. Overall, our results demonstrate that left FEF, a high-level region along the auditory “where” pathway, encodes sound locations during passive listening in two ways: a univariate opponent hemifield activation representation and a multivariate full-field activation pattern representation.

BackgroundNon-motor symptoms are common in Parkinson's disease (PD) patients, decreasing quality of life and having no specific treatments. This research investigates dynamic functional connectivity (FC) changes during PD duration and its correlations with non-motor symptoms. MethodsTwenty PD patients and 19 healthy controls (HC) from PPMI dataset were collected and used in this study. Independent component analysis (ICA) was performed to select significant components from the entire brain. Components were grouped into seven resting-state intrinsic networks. Static and dynamic FC changes during resting-state functional magnetic resonance imaging (fMRI) were calculated based on selected components and resting state networks (RSN). ResultsStatic FC analysis results showed that there was no difference between PD-baseline (PD-BL) and HC group. Network averaged connection between frontoparietal network and sensorimotor network (SMN) of PD-follow up (PD-FU) was lower than PD-BL. Dynamic FC analysis results suggested four distinct states, and each state's temporal characteristics, such as fractional windows and mean dwell time, were calculated. The state 2 of our study showed positive coupling within and between SMN and visual network, while the state 3 showed hypo-coupling through all RSN. The fractional windows and mean dwell time of PD-FU state 2 (positive coupling state) were statistically lower than PD-BL. Fractional windows and mean dwell time of PD-FU state 3 (hypo-coupling state) were statistically higher than PD-BL. Outcome scales in Parkinson's disease-autonomic dysfunction scores of PD-FU positively correlated with mean dwell time of state 3 of PD-FU. ConclusionOverall, our finding indicated that PD-FU patients spent more time in hypo-coupling state than PD-BL. The increase of hypo-coupling state and decrease of positive coupling state might correlate with the worsening of non-motor symptoms in PD patients. Dynamic FC analysis of resting-state fMRI can be used as monitoring tool for PD progression.

目的研究言语康复训练改变唇腭裂(cleft lip and palate,CLP)患者静息态脑网络的时间动态功能连接的脑机制.资料与方法经纳入和排除标准筛选后选取未接受言语康复训练及接受言语康复训练且合格的CLP患者各22例,对所有患者进行汉语语音清晰度测试字表评估和功能磁共振成像(functional magnetic resonance imaging,fMRI)检查.使用静息态fMRI,通过独立成分分析进行动态功能网络连接,比较两组受试者脑网络及动态时间特性的组间差异.结果我们选取了21个符合要求的独立成分,并确定了动态功能连接模式的两种状态.状态1中,与言语康复训练之前的CLP患者相比,言语康复之后的CLP患者的视觉网络和执行控制网络的功能连接减弱(p<0.05),小脑网络和视觉网络、听觉网络、感觉运动网络的功能连接均增强(p<0.05);状态2中,言语康复训练之后的CLP患者的额顶网络与视觉网络、默认模式网络、感觉运动网络的功能连接均加强(p<0.05),视觉网络与执行控制网络、感觉运动网络的功能连接均减弱(p<0.05).两组受试者在动态功能连接分析中的时间特性参数均无统计学差异,但言语康复训练之前的CLP患者在状态2中的平均驻留时间较短,同时该组的状态转换次数更多.结论CLP患者动态脑网络之间功能连接的恢复和时间特性的改变为言语康复训练的脑机制提供了客观解释,并为进一步评判患者言语康复情况提供了参考.

Epicardial adipose tissue (EAT) is contiguous with arteries and myocardium. An increase in the volume of EAT may lead to adverse cardiovascular events. Therefore, quantification of EAT is necessary. The purpose of this paper is to employ a more than helpful algorithm for EAT segmentation and quantification. First, we used a simple convolutional neural network to select EAT slices, which significantly reduced oversegmentation. Then, we employed multiscale residual attention Unet (MRA-Unet) to achieve EAT segmentation based on the selected slices. Finally, we calculated the segmented volume to quantify EAT. We used 33/103 patients to test the model. The average Dice score for EAT segmentation was 0.883. For EAT quantification, the Pearson and concordance correlation coefficients reached 0.973 and 0.971, respectively. The results showed that our algorithm had strong agreement and consistency with expert. Our method performed efficient quantification and had strong consistency and agreement with the volume manually marked by experts. This algorithm can be used as a tool to assist in the clinical quantification of EAT. By combining different measurements to predict adverse cardiovascular and heart disease events, it has the potential to be applied for clinical use in the future.

Abstract Our ability to localize low-frequency sounds relies on interaural time differences (ITD), the sensitivity to which underlies our ability to localize sound sources in azimuth and to segregate competing sound sources across a wide range of situations. However, the cortical representation of ITD has not been well characterized. To investigate differences in evoked cortical potentials and global functional connectivity during the unattended processing of sound localization on the horizontal plane, auditory mismatch negativity (MMN) was recorded and analyzed in adults using a deviant-standard oddball paradigm. A central sound source at the centerline of the horizontal plane (ITD = 0μs) was used as the standard stimulus and lateralized sounds with various ITDs were used as deviant stimuli. Characteristics of event-related potentials (ERP) and MMN obtained from young adults were compared between different deviant stimulus groups. The time-frequency (TF) power at each electrode associated with standard and deviant conditions were calculated. Distribution differences of electrode power between standard and deviant stimuli among oddball experiments were compared. Phase lag index (PLI) was also calculated to examine the dynamics of functional connectivity in various deviant stimulus groups. Network topological parameters were calculated to examine the global efficiency of information transfer in different deviant stimulus groups. Results of MMN analysis indicated that sounds with larger deviations from centerline elicited MMNs with greater amplitudes and shorter latencies and the MMN response was stronger in the hemisphere contralateral to the deviant stimulus. Results from the TF analysis showed a significant event-related synchronization (ERS) in the theta frequency band. Results of PLI functional connectivity suggested that the deviant stimulus resulted in a global increase in connectivity in theta band. Increased functional connectivity was found in frontal electrode pairs in sound stimuli towards the central rather than peripheral sound stimuli, and the peripheral sound stimuli revealed a higher global efficiency. This study suggests that the auditory system response in horizontal sound localization appeared to originate from the activity of the theta band. Sound sources with larger deviations from the centerline revealed higher brain network efficiency.

Cleft lip and/or palate (CLP) are the most common craniofacial malformations in humans. Speech problems often persist even after cleft repair, such that follow-up articulation training is usually required. However, the neural mechanism behind effective articulation training remains largely unknown. We used fMRI to investigate the differences in brain activation, functional connectivity, and effective connectivity across CLP patients with and without articulation training and matched normal participants. We found that training promoted task-related brain activation among the articulation-related brain networks, as well as the global attributes and nodal efficiency in the functional-connectivity-based graph of the network. Our results reveal the neural correlates of effective articulation training in CLP patients, and this could contribute to the future improvement of the post-repair articulation training program.

随着大数据、互联网、信息科技和人工智能技术的飞速发展,智慧医疗、"大健康"、医疗大数据等概念被提出并得到国内外医疗健康及教育等多领域的高度关注.民生健康在人类及国家发展进步中战略性地位的提升也大力促进了智慧医疗行业的高速发展.20世纪70年代,国外最早出现人工智能技术在医疗健康领域的尝试,20世纪80年代我国开始进行智慧医疗的开发研究.到21世纪初,智慧医疗行业已经取得非凡成绩,在智慧医院、远程医疗、肿瘤智能化辅助诊疗、眼部疾病早期预测、健康管理等领域取得了突破性成就,为进一步实现安全、高效、全面、智能化的全民医疗保障系统,促进规范化标准医疗健康大数据库构建,推动智慧医疗的发展和优化奠定了基础.本文概括总结了近年来智慧医疗在国内外的发展和应用现状,介绍了智慧医疗实际应用中的智能识别、信息融合、移动计算、云计算等关键技术,指出了智慧医疗的未来发展方向.