RESUMEN
Increasing evidence suggests that cortical folding patterns of human cerebral cortex manifest overt structural and functional differences. However, for interpretability, few studies leverage advanced techniques (e.g., deep learning) to investigate the difference among cortical folds, resulting in more differences yet to be extensively explored. To this end, we proposed an effective topology-preserving transfer learning framework to differentiate cortical fMRI time series extracted from cortical folds. Our framework consists of three main parts: (1) Neural architecture search (NAS), which is used to devise a well-performing network structure based on an initialized hand-designed super-graph in an image dataset; (2) Topology-preserving transfer, which takes the model searched by NAS as the source network, keeping the topological connectivity in the network unchanged, while transforming all 2D operations including convolution and pooling into 1D, therefore resulting in a topology-preserving network, named TPNAS-Net; (3) Classification and correlation analysis, which involves using the TPNAS-Net to classify 1D cortical fMRI time series for each individual brain, and performing a group difference analysis between autism spectrum disorder (ASD) and healthy control (HC) and correlation analysis with clinical information (i.e., age). Extensive experiments on two ASD datasets obtain consistent results, demonstrating that the TPNAS-Net not only discriminates cortical folding patterns at high classification accuracy, but also captures subtle differences between ASD and HC (p-value = 0.042). In addition, we discover that there is a positive correlation between the classification accuracy and age in ASD (r = 0.39, p-value = 0.04). These findings together suggest that structural and functional differences in cortical folding patterns between ASD and HC may provide a potentially useful biomarker for the diagnosis of ASD.
Asunto(s)
Trastorno del Espectro Autista , Encéfalo , Corteza Cerebral/diagnóstico por imagen , Humanos , Aprendizaje Automático , Imagen por Resonancia Magnética , Extractos VegetalesRESUMEN
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.
Asunto(s)
Conectoma , Encéfalo/diagnóstico por imagen , Humanos , Imagen por Resonancia MagnéticaRESUMEN
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.
Asunto(s)
Imagen por Resonancia Magnética/métodos , Redes Neurales de la Computación , Encéfalo/diagnóstico por imagen , HumanosRESUMEN
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.
Asunto(s)
Encéfalo/diagnóstico por imagen , Conectoma/métodos , Aprendizaje Profundo , Imagen por Resonancia Magnética/métodos , Encéfalo/fisiología , Emociones/fisiología , Humanos , Lenguaje , Vías Nerviosas/diagnóstico por imagen , Vías Nerviosas/fisiología , Análisis Espacio-TemporalRESUMEN
Mapping the relation between cortical convolution and structural/functional brain architectures could provide deep insights into the mechanisms of brain development, evolution and diseases. In our previous studies, we found a unique gyral folding pattern, termed a 3-hinge, which was defined as the conjunction of three gyral crests. The uniqueness of the 3-hinge was evidenced by its thicker cortex and stronger fiber connections than other gyral regions. However, the role that 3-hinges play in cortico-cortical connective architecture remains unclear. To this end, we conducted MRI studies by constructing structural cortico-cortical connective networks based on a fine-granular cortical parcellation, the parcels of which were automatically labeled as 3-hinge, 2-hinge (ordinary gyrus) or sulcus. On human brains, 3-hinges possess significantly higher degrees, strengths and betweennesses than 2-hinges, suggesting that 3-hinges could serve more like hubs in the cortico-cortical connective network. This hypothesis gains supports from human functional network analyses, in which 3-hinges are involved in more global functional networks than ordinary gyri. In addition, 3-hinges could serve as 'connector' hubs rather than 'provincial' hubs and they account for a dominant proportion of nodes in the high-level 'backbone' of the network. These structural results are reproduced on chimpanzee and macaque brains, while the roles of 3-hinges as hubs become more pronounced in higher order primates. Our new findings could provide a new window to the relation between cortical convolution, anatomical connection and brain function.
Asunto(s)
Conectoma , Animales , Encéfalo/diagnóstico por imagen , Mapeo Encefálico , Macaca , Imagen por Resonancia Magnética , Red Nerviosa/diagnóstico por imagen , Vías Nerviosas/diagnóstico por imagenRESUMEN
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.
Asunto(s)
Encéfalo , Conectoma , Encéfalo/diagnóstico por imagen , Humanos , Imagen por Resonancia Magnética , Redes Neurales de la Computación , Programas InformáticosRESUMEN
The human cerebral cortex is highly folded into diverse gyri and sulci. Accumulating evidences suggest that gyri and sulci exhibit anatomical, morphological, and connectional differences. Inspired by these evidences, we performed a series of experiments to explore the frequency-specific differences between gyral and sulcal neural activities from resting-state and task-based functional magnetic resonance imaging (fMRI) data. Specifically, we designed a convolutional neural network (CNN) based classifier, which can differentiate gyral and sulcal fMRI signals with reasonable accuracies. Further investigations of learned CNN models imply that sulcal fMRI signals are more diverse and more high frequency than gyral signals, suggesting that gyri and sulci truly play different functional roles. These differences are significantly associated with axonal fiber wiring and cortical thickness patterns, suggesting that these differences might be deeply rooted in their structural and cellular underpinnings. Further wavelet entropy analyses demonstrated the validity of CNN-based findings. In general, our collective observations support a new concept that the cerebral cortex is bisectionally segregated into 2 functionally different units of gyri and sulci.
Asunto(s)
Corteza Cerebral/anatomía & histología , Corteza Cerebral/fisiología , Adulto , Mapeo Encefálico/métodos , Humanos , Procesamiento de Imagen Asistido por Computador , Imagen por Resonancia Magnética , Redes Neurales de la Computación , Adulto JovenRESUMEN
Comparison and integration of neuroimaging data from different brains and populations are fundamental in neuroscience. Over the past decades, the neuroimaging field has largely depended on image registration to compare and integrate neuroimaging data from individuals in a common reference space, with a basic assumption that the brains are similar. However, the intrinsic neuroanatomical complexity and huge interindividual cortical folding variation remain underexplored. Here we focus on a specific cortical convolution pattern, termed 3-hinge gyral folding, which is the conjunction of gyri from multiple orientations and has unique and consistent anatomically, structurally, and functionally connective patterns across subjects. By developing a novel shape descriptor and a two-stage clustering pipeline, we devise an automatic method to identify 3-hinges in the Human Connectome Project Q3 868 human brains, and further parameterize the complexity of such a pattern and quantify its regularity and variation in terms of 3-hinge number, position, and morphology. Our results not only exhibit the huge interindividual variations, but also reveal regular relationship between gyral hinges and other factors, such as their locations and cortical morphologies. It is found that "line-shape" cortices have relatively more consistent 3-hinge shape pattern distributions, and certain types of 3-hinge patterns favor particular cortical morphologies. In addition, more 3-hinges are found on "line-shape" cortices while their numbers vary more across subjects than those on "non-line-shape" cortices. This study adds new insights into a better understanding of the regularity and variability of human brain anatomy, and their functional aspects.
Asunto(s)
Corteza Cerebral/anatomía & histología , Conectoma , Interpretación Estadística de Datos , Procesamiento de Imagen Asistido por Computador/métodos , Imagen por Resonancia Magnética/métodos , Adulto , Corteza Cerebral/diagnóstico por imagen , HumanosRESUMEN
fMRI data decomposition techniques have advanced significantly from shallow models such as Independent Component Analysis (ICA) and Sparse Coding and Dictionary Learning (SCDL) to deep learning models such Deep Belief Networks (DBN) and Convolutional Autoencoder (DCAE). However, interpretations of those decomposed networks are still open questions due to the lack of functional brain atlases, no correspondence across decomposed or reconstructed networks across different subjects, and significant individual variabilities. Recent studies showed that deep learning, especially deep convolutional neural networks (CNN), has extraordinary ability of accommodating spatial object patterns, e.g., our recent works using 3D CNN for fMRI-derived network classifications achieved high accuracy with a remarkable tolerance for mistakenly labelled training brain networks. However, the training data preparation is one of the biggest obstacles in these supervised deep learning models for functional brain network map recognitions, since manual labelling requires tedious and time-consuming labours which will sometimes even introduce label mistakes. Especially for mapping functional networks in large scale datasets such as hundreds of thousands of brain networks used in this paper, the manual labelling method will become almost infeasible. In response, in this work, we tackled both the network recognition and training data labelling tasks by proposing a new iteratively optimized deep learning CNN (IO-CNN) framework with an automatic weak label initialization, which enables the functional brain networks recognition task to a fully automatic large-scale classification procedure. Our extensive experiments based on ABIDE-II 1099 brains' fMRI data showed the great promise of our IO-CNN framework.
Asunto(s)
Trastorno Autístico/diagnóstico por imagen , Mapeo Encefálico/métodos , Procesamiento de Imagen Asistido por Computador/métodos , Imagen por Resonancia Magnética/métodos , Redes Neurales de la Computación , Adolescente , Adulto , Algoritmos , Niño , Preescolar , Femenino , Humanos , Masculino , Persona de Mediana Edad , Programas InformáticosRESUMEN
Recent studies have shown that quantitative description of gyral shape patterns offers a novel window to examine the relationship between brain structure and function. Along this research line, this paper examines a unique and interesting type of cortical gyral region where 3 different gyral crests meet, termed 3-hinge gyral region. We extracted 3-hinge gyral regions in macaque/chimpanzee/human brains, quantified and compared the relevant DTI-derived fiber densities in 3-hinge and 2-hinge gyral regions. Our observations consistently showed that DTI-derived fiber densities in 3-hinge regions are much higher than those in 2-hinge regions. Therefore, we hypothesize that besides the cortical expansion, denser fiber connections can induce the formation of 3-hinge gyri. To examine the biomechanical basis of this hypothesis, we constructed a series of 3-dimensional finite element soft tissue models based on continuum growth theory to investigate fundamental biomechanical mechanisms of consistent 3-hinge gyri formation. Our computational simulation results consistently showed that during gyrification gyral regions with higher concentrations of growing axonal fibers tend to form 3-hinge gyri. Our integrative approach combining neuroimaging data analysis and computational modeling appears effective in probing a plausible theory of 3-hinge gyri formation and providing new insights into structural and functional cortical architectures and their relationship.
Asunto(s)
Mapeo Encefálico , Encéfalo/diagnóstico por imagen , Encéfalo/fisiología , Simulación por Computador , Modelos Neurológicos , Fibras Nerviosas/fisiología , Adulto , Algoritmos , Animales , Imagen de Difusión por Resonancia Magnética , Humanos , Macaca , Pan troglodytes , Especificidad de la Especie , Adulto JovenRESUMEN
One distinct feature of the cerebral cortex is its convex (gyri) and concave (sulci) folding patterns. Due to the remarkable complexity and variability of gyral/sulcal shapes, it has been challenging to quantitatively model their organization patterns. Inspired by the observation that the lines of gyral crests can form a connected graph on each brain hemisphere, we propose a new representation of cortical gyri/sulci organization pattern - gyral net, which models cortical architecture from a graph perspective, starting with nodes and edges obtained from the reconstructed cortical surfaces. A novel computational framework is developed to efficiently and automatically construct gyral nets from surface meshes, and four measurements are devised to quantify the folding patterns. Using an MRI dataset for autism study as a test bed, we identified reduced local connectivity cost and increased curviness of gyral net bilaterally on the parietal lobe, occipital lobe, and temporal lobe in autistic patients. Additionally, we found that the cortical thickness and the gyral straightness of gyral joints are higher than the rest of gyral crest regions. The proposed representation offers a new tool for a comprehensive and reliable characterization of the cortical folding organization. This novel computational framework will enable large-scale analyses of cortical folding patterns in the future.