Multi-Modal Non-Euclidean Brain Network Analysis With Community Detection and Convolutional Autoencoder

Brain network analysis has been proved to be one of the most effective methods in brain disease diagnosis. In order to construct discriminative brain networks and improve the performance of disease diagnosis, many machine learning–based methods have been proposed. Recent studies show that combining functional and structural brain networks is more effective than using only single modality data. However, in the most of existing multi-modal brain network analysis methods, it is a common strategy that constructs functional and structural network separately, which is difficult to embed complementary information of different modalities of brain network. To address this issue, we propose a unified brain network construction algorithm, which jointly learns both functional and structural data and effectively face the connectivity and node features for improving classification. First, we conduct space alignment and brain network construction under a unified framework, and then build the correlation model among all brain regions with functional data by low-rank representation so that the global brain region correlation can be captured. Simultaneously, the local manifold with structural data is embedded into this model to preserve the local structural information. Second, the PageRank algorithm is adaptively used to evaluate the significance of different brain regions, in which the interaction of multiple brain regions is considered. Finally, a multi-kernel strategy is utilized to solve the data heterogeneity problem and merge the connectivity as well as node information for classification. We apply the proposed method to the diagnosis of epilepsy, and the experimental results show that our method can achieve a promising performance.

The fusion of structural and functional brain network analysis has been widely applied in the analysis of various brain diseases, especially in identifying the progression stages of Alzheimer’s disease. However, most existing multimodal brain network analysis methods separately construct functional and structural networks, making it difficult to incorporate complementary information from several modalities. To tackle this difficulty, we propose AMFusion(Adaptive Multimodal Fusion), a unified brain network construction framework that jointly learns from functional and structural images, thereby efficiently resolving connectivity and node feature learning problems. Our approach begins with a designed quantization encoder to extract structural features from DTI, while dynamic functional connectivity (FC) is constructed from fMRI. Subsequently, a multilevel brain network fusion module is employed to learn and integrate brain connections, considering interactions across different temporal and spatial scales. Finally, a classifier guides further optimization of the brain network, facilitating multi-class disease classification. The performance of the proposed AMFusion was validated using the authentic dataset from ADNI, and experimental results demonstrated the effectiveness of AMFusion. By considering both connection patterns and node features, our method overcomes the limitations of existing approaches, providing a more comprehensive and robust framework for analyzing and diagnosing AD.

Recent advancements in AI, big data analytics, and magnetic resonance imaging (MRI) have revolutionized the study of brain diseases such as Alzheimer's Disease (AD). However, most AI models used for neuroimaging classification tasks have limitations in their learning strategies, that is batch training without the incremental learning capability. To address such limitations, the systematic Brain Informatics methodology is reconsidered to realize evidence combination and fusion computing with multi-modal neuroimaging data through continuous learning. Specifically, we introduce the BNLoop-GAN (Loop-based Generative Adversarial Network for Brain Network) model, utilizing multiple techniques such as conditional generation, patch-based discrimination, and Wasserstein gradient penalty to learn the implicit distribution of brain networks. Moreover, a multiple-loop-learning algorithm is developed to combine evidence with better sample contribution ranking during training processes. The effectiveness of our approach is demonstrated through a case study on the classification of individuals with AD and healthy control groups using various experimental design strategies and multi-modal brain networks. The BNLoop-GAN model with multi-modal brain networks and multiple-loop-learning can improve classification performance.

Adolescent MDD has become a significant public health issue, yet its underlying mechanisms remain unclear. Multimodal brain imaging techniques offer a powerful method for exploring complex mental disorders. However, evidence focusing on the multimodal brain networks and structural-functional coupling in adolescent depression is still limited. Participants with major depressive disorder (MDD) were Han Chinese individuals aged 13 to 18 who had been unmedicated for at least two weeks. We conducted multimodal MRI acquisitions, including structural MRI (sMRI), resting-state functional MRI (rsfMRI), and Diffusion Tensor Imaging (DTI). The cortex was parceled into 360 regions using the HCP-MMP atlas. Functional connectivity and deterministic structural connectivity matrices were constructed, and structural-functional coupling coefficients were calculated. Differences in connectivity and coupling coefficients between the MDD and healthy controls (HCs) groups were identified. A total of 25 adolescents with MDD (mean age: 15.68 years, standard deviation [SD]: 1.18; Female: 21 (84.00%)) and 27hCs (mean age: 14.30 years, standard deviation [SD]: 1.51; Female: 13 (48.15%)) were included in the analysis. There were 9 structural connections and 122 functional connections that differed between the two groups, involving multiple cortical regions. Additionally, we identified structural-functional coupling differences in three brain areas, specifically the posterior cingulate cortex and the ventral visual cortex. Adolescent MDD involves disruptions in brain structural networks, functional networks, and structural-functional coupling. These differing indicators may serve as potential biomarkers for adolescent MDD.

Learning from the multimodal brain imaging data attracts a large amount of attention in medical image analysis due to the proliferation of multimodal data collection. It is widely accepted that multimodal data can provide complementary information than mining from a single modality. However, unifying the image-based knowledge from the multimodal data is very challenging due to different image signals, resolution, data structure, etc.. In this study, we design a supervised deep model to jointly analyze brain morphometry and functional connectivity on the cortical surface and we name it deep multimodal brain network learning (DMBNL). Two graph-based kernels, i.e., geometry-aware surface kernel (GSK) and topology-aware network kernel (TNK), are proposed for processing the cortical surface morphometry and brain functional network. The vertex features on the cortical surface from GSK is pooled and feed into TNK as its initial regional features. In the end, the graph-level feature is computed for each individual and thus can be applied for classification tasks. We test our model on a large autism imaging dataset. The experimental results prove the effectiveness of our model.

Patients with Parkinson's disease (PD) present various responsiveness to levodopa, but the cause of such differences in levodopa responsiveness is unclear.Previous studies related the damage of brain white matter (WM) to levodopa responsiveness in PD patients, but no study investigated the relationship between the structural brain network change in PD patients and their levodopa responsiveness. PD patients were recruited and evaluated using the Unified Parkinson's Disease Rating Scale (UPDRS). Each patient received a diffusion tensor imaging (DTI) scan and an acute levodopa challenge test. The improvement rate of UPDRS-III was calculated. PD patients were grouped into irresponsive group (improvement rate<30%) and responsive group (improvement rate ≥ 30%). Tract-based spatial statistics (TBSS), deterministic tracing (DT), region of interest (ROI) analysis, and automatic fiber identification (AFQ) analyses were performed.The structural brain network was also constructed and the topological parameters were calculated. Fifty-four PD patients were included. TBSS identified significant differences in fractional anisotropy (FA) values in the corpus callosum and other regions of the brain. DT and ROI analysis of the corpus callosum found a significant difference in FA between the two groups. Graph theory analysis showed statistical differences in global efficiency, local efficiency, and characteristic path length. PD patients with poor responsiveness to levodopa had WM damage in multiple brain areas, especially the corpus callosum, which might cause disruption of information integration of the structural brain network.

Dynamic evidence fusion neural networks with uncertainty theory and its application in brain network analysis

Local structural-functional coupling with counterfactual explanations for epilepsy prediction.

Network science has become an invaluable tool for neuroimaging analysis and has advanced our understanding of the brain's complex structural and functional topology. Network science, an interdisciplinary offshoot of graph theory, conceives of the brain as a system of nodes, representing brain regions or voxels, and edges, representing the structural or functional connection between these regions or voxels. Such a model is considered appealing in neuroscience as it describes a system with various interacting regions that produces complex behaviors. This conception of the brain has revealed that the brain exhibits “small-world” properties, namely that brain networks are systems with regional specialization and efficient global information transfer. More importantly, studies in brain networks have furthered our understanding of diseases and disorders affecting the brain. Assessing group differences in representative samples of brain networks has become increasingly important, yet appropriate methodological developments have remained relatively sparse. Additionally, there is increased interest in understanding dynamic changes, or how a network changes over time. Currently, most network analysis tools are designed to analyze a static representation, thus changes that occur over time are often overlooked or ignored. As the field continues to grow, it is becoming increasingly important to develop methods for understanding the complexity of these data. This research topic features selected contributions that explores group analysis and dynamics in brain network studies including methodological and practical applications employing various imaging modalities. This snapshot of the field represents the growing interest to expand and validate new methods for use in the field of brain network neuroscience. “Defining nodes in complex brain networks” is a review article discussing the methodological choices for defining nodes, particularly how the choice of atlas-based or voxel-based methods affects the computed network (Stanley et al., 2013). “Routing in the brain” is a commentary on network routing focusing on how current methods and models may not be adequate for brain networks. Further exploration of the packet switching network model may be more realistic and appropriate for the brain (Graham, 2014). “A multimodal approach for determining brain networks by jointly modeling functional and structural connectivity” is a methods article introducing a Bayesian model for estimating functional networks, from functional magnetic resonance imaging (fMRI) data, leveraging complementary structural diffusion tensor imaging (DTI) data (Xue et al., 2015). “A permutation testing framework to compare groups of brain networks” is a methods article describing a permutation testing framework to statistically compare groups of functional brain networks while incorporating topological features inherent in each individual network (Simpson et al., 2013). “Assessing dynamics, spatial scale, and uncertainty in task-related brain network analyses” is a methods article describing a statistical approach for computing uncertainty in static and dynamic functional networks and aggregating network measures in task-related electrocorticography (ECoG) data (Stephen et al., 2014). “Hierarchical vector auto-regressive models and their applications to multi-subject effective connectivity” is a methods article introducing a generalization of the vector auto-regressive (VAR) model for comparing effective connectivity between experimental conditions while accounting for between-subject heterogeneity (Gorrostieta et al., 2013). “Statistical network analysis for functional MRI: summary networks and group comparisons” is a review article discussing the construction of summary networks and how to test for topological differences in groups of networks when these groups also exhibit significant differences in density (Ginestet et al., 2014). “Detecting functional connectivity change points for single-subject fMRI data” is a methods article that extends Dynamic Connectivity Regression (DCR), a data-driven technique which detects change points in functional connectivity between brain regions to determine state changes over the course of an experimental task, by introducing a novel algorithm aimed at increasing the estimation of networks for individual subject data (Cribben et al., 2013). “The Laplacian spectrum of neural networks” is an original research article examining the use of the Laplacian spectrum in anatomic networks of the macaque, cat and Caenorhabditis elegans. This method describes network structure at a systems level, assessing systemic infrastructural properties of the entire network as opposed to properties of specific nodes or connections (De Lange et al., 2014). “Quantifying network properties in multi-electrode recordings: spatiotemporal characterization and inter-trial variation of evoked gamma oscillations in mouse somatosensory cortex in vitro” is an original research article on network dynamics in multi-electrode array data focusing on changes in the functional connectivity of evoked gamma oscillations in cortical circuits (Carmeli et al., 2013).

Many studies, using a variety of imaging techniques, have shown that deafness induces functional plasticity in the brain of adults with late-onset deafness, and in children changes the way the auditory brain develops. Cross modal plasticity refers to evidence that stimuli of one modality (e.g. vision) activate neural regions devoted to a different modality (e.g. hearing) that are not normally activated by those stimuli. Other studies have shown that multimodal brain networks (such as those involved in language comprehension, and the default mode network) are altered by deafness, as evidenced by changes in patterns of activation or connectivity within the networks. In this paper, we summarise what is already known about brain plasticity due to deafness and propose that functional near-infra-red spectroscopy (fNIRS) is an imaging method that has potential to provide prognostic and diagnostic information for cochlear implant users. Currently, patient history factors account for only 10 % of the variation in post-implantation speech understanding, and very few post-implantation behavioural measures of hearing ability correlate with speech understanding. As a non-invasive, inexpensive and user-friendly imaging method, fNIRS provides an opportunity to study both pre- and post-implantation brain function. Here, we explain the principle of fNIRS measurements and illustrate its use in studying brain network connectivity and function with example data.

Graph neural networks for fMRI functional brain networks: A survey.

Fundamentals of Brain Network Analysis

The integration of artificial intelligence with advanced neuroimaging techniques represents one of the most important developments in contemporary neuroscience and clinical neuroanatomy. This narrative review summarizes how machine learning and deep learning methods can be applied to neuroimaging data to reveal hidden anatomical and functional brain patterns with direct clinical relevance. Neuroanatomy plays a central role in this process, providing the structural framework for interpreting functional, metabolic, and electrophysiological signals. Artificial intelligence-based analysis enables the detection of subtle alterations in gray and white matter, the analysis of brain networks and pathways, and the integration of multimodal imaging data beyond the limits of human perception and conventional statistical approaches. In addition to aiding physicians in establishing the correct diagnosis, these capabilities contribute to earlier and more accurate diagnoses, improved disease differentiation, identification of affected structures within specific health conditions or diseases, and more reliable prognostic assessment. In addition to diagnostic applications, artificial intelligence supports the development of personalized therapeutic strategies, including targeted neuromodulation, neuroimaging-guided surgery, adaptive brain-computer interfaces, and an individualized approach in further disease monitoring. Despite existing challenges related to data heterogeneity, model interpretability, and ethical regulation, the synergy between AI and neuroimaging represents a critical step toward personalized medicine. By leveraging detailed neuroanatomical information, artificial intelligence-driven approaches enable therapies tailored to the unique brain structure and network organization of each individual.

Background: Changes in the metabolic and structural brain networks in mild cognitive impairment (MCI) have been widely researched. However, few studies have compared the differences in the topological properties of the metabolic and structural brain networks in patients with MCI.Methods: We analyzedmagnetic resonance imaging (MRI) and fluoro-deoxyglucose positron emission tomography (FDG-PET) data of 137 patients with MCI and 80 healthy controls (HCs). The HC group data comes from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. The permutation test was used to compare the network parameters (characteristic path length, clustering coefficient, local efficiency, and global efficiency) between the two groups. Partial Pearson’s correlation analysis was used to calculate the correlations of the changes in gray matter volume and glucose intake in the key brain regions in MCI with the Alzheimer’s Disease Assessment Scale-Cognitive (ADAS-cog) sub-item scores.Results: Significant changes in the brain network parameters (longer characteristic path length, larger clustering coefficient, and lower local efficiency and global efficiency) were greater in the structural network than in the metabolic network (longer characteristic path length) in MCI patients than in HCs. We obtained the key brain regions (left globus pallidus, right calcarine fissure and its surrounding cortex, left lingual gyrus) by scanning the hubs. The volume of gray matter atrophy in the left globus pallidus was significantly positively correlated with comprehension of spoken language (p = 0.024) and word-finding difficulty in spontaneous speech item scores (p = 0.007) in the ADAS-cog. Glucose intake in the three key brain regions was significantly negatively correlated with remembering test instructions items in ADAS-cog (p = 0.020, p = 0.014, and p = 0.008, respectively).Conclusion: Structural brain networks showed more changes than metabolic brain networks in patients with MCI. Some brain regions with significant changes in betweenness centrality in both structural and metabolic networks were associated with MCI.

BackgroundRecently, many academic research groups focused their attention on changes in human brain networks related to several kinds of pathologies and diseases. Generally speaking, Network Medicine promises to identify the principles to understand, at the molecular level, the human life complex system. We applied the theoretical approach of the Network Medicine mainly to brain diseases, testing the potentiality of the method to produce an early and more in-depth diagnostic process. However, the range of our topics and application of the Network Medicine spans from common neurological and neuropsychological spectrum disorders to other physiological systems or district, only apparently out of the action range of the neurological one. In the areas of Neurology and Neuropsychology we are studying brain networks in Tourette Syndrome [TS] and Multiple Sclerosis [MS]. Moreover, we are studying brain networking related to genotype alteration responsible for different gustatory stimuli processing in the brain. Further, we recently started to study infective diseases and immunological system.Main body of the abstractThe analysis of brain networks is made feasible by the development and application of imaging acquisition methods based on the Magnetic Resonance Imaging and functional MRI, as well as the availability of innovative calculation tools from graph theory and complex dynamical systems. This paper highlights the Network Medicine concepts and the application of the functional connectivity for the brain networks description. We will describe all these methods with a reduced use of formulas, along with some technical tools available in the WEB to perform the calculations of the related parameters. We will briefly describe the freely available software CONN used for calculating different connectivity parameters. In all these studies, the networks characterisation is performed using indexes from metric and topology of the brain networks. The possibility of early recognition of the diseases by the identification of the alteration in the network’s parameters may significantly improve patient outcome, with also significant social community benefits. We intend to illustrate a possible pipeline for the Networks Brain application to neuroradiologists involved in the diagnosis of complex diseases.Short conclusionThe multidisciplinary approach is one of the essential characteristics of journals in the imaging sciences field; the issues and the methodologies examined in this paper could be of great interest to the reader’s community. Furthermore, until today very few studies focused on the network medicine and connectivity explored in functional MRI. Hence this might have an additional value for the journal.