Unveiling the neuromorphological space

Multi-level feature fusion network for neuronal morphology classification

Neuronal morphology can be represented using various feature representations, such as hand-crafted morphometrics and deep features. These features are complementary to each other, contributing to improving performance. However, existing classification methods only utilize a single feature representation or simply concatenate different features without fully considering their complementarity. Therefore, their performance is limited and can be further improved. In this paper, we propose a multi-level feature fusion network that fully utilizes diverse feature representations and their complementarity to effectively describe neuronal morphology and improve performance. Specifically, we devise a Multi-Level Fusion Module (MLFM) and incorporate it into each feature extraction block. It can facilitate the interaction between different features and achieve effective feature fusion at multiple levels. The MLFM comprises a channel attention-based Feature Enhancement Module (FEM) and a cross-attention-based Feature Interaction Module (FIM). The FEM is used to enhance robust morphological feature presentations, while the FIM mines and propagates complementary information across different feature presentations. In this way, our feature fusion network ultimately yields a more distinctive neuronal morphology descriptor that can effectively characterize neurons than any singular morphological representation. Experimental results show that our method effectively depicts neuronal morphology and correctly classifies 10-type neurons on the NeuronMorpho-10 dataset with an accuracy of 95.18%, outperforming other approaches. Moreover, our method performs well on the NeuronMorpho-12 and NeuronMorpho-17 datasets and possesses good generalization.

Graph Representation Learning for Large-Scale Neuronal Morphological Analysis

The analysis of neuronal morphological data is essential to investigate the neuronal properties and brain mechanisms. The complex morphologies, absence of annotations, and sheer volume of these data pose significant challenges in neuronal morphological analysis, such as identifying neuron types and large-scale neuron retrieval, all of which require accurate measuring and efficient matching algorithms. Recently, many studies have been conducted to describe neuronal morphologies quantitatively using predefined measurements. However, hand-crafted features are usually inadequate for distinguishing fine-grained differences among massive neurons. In this article, we propose a novel morphology-aware contrastive graph neural network (MACGNN) for unsupervised neuronal morphological representation learning. To improve the retrieval efficiency in large-scale neuronal morphological datasets, we further propose Hash-MACGNN by introducing an improved deep hash algorithm to train the network end-to-end to learn binary hash representations of neurons. We conduct extensive experiments on the largest dataset, NeuroMorpho, which contains more than 100 000 neurons. The experimental results demonstrate the effectiveness and superiority of our MACGNN and Hash-MACGNN for large-scale neuronal morphological analysis.

Retrieving similar substructures on 3D neuron reconstructions

Since manual tracing is time consuming and the performance of automatic tracing is unstable, it is still a challenging task to generate accurate neuron reconstruction efficiently and effectively. One strategy is generating a reconstruction automatically and then amending its inaccurate parts manually. Aiming at finding inaccurate substructures efficiently, we propose a pipeline to retrieve similar substructures on one or more neuron reconstructions, which are very similar to a marked problematic substructure. The pipeline consists of four steps: getting a marked substructure, constructing a query substructure, generating candidate substructures and retrieving most similar substructures. The retrieval procedure was tested on 163 gold standard reconstructions provided by the BigNeuron project and a reconstruction of a mouse’s large neuron. Experimental results showed that the implementation of the proposed methods is very efficient and all retrieved substructures are very similar to the marked one in numbers of nodes and branches, and degree of curvature.

Neuronal Morphology and Synapse Count in the Nematode Worm

The somatic nervous system of the nematode worm Caenorhabditis elegans is a model for understanding the physical characteristics of the neurons and their interconnections. Its neurons show high variation in morphological attributes. This study investigates the relationship of neuronal morphology to the number of synapses per neuron. Morphology is also examined for any detectable association with neuron cell type or ganglion membership.

Pinpointing Morphology and Projection of Excitatory Neurons in Mouse Visual Cortex

The excitatory neurons in the visual cortex are of great significance for us in understanding brain functions. However, the diverse neuron types and their morphological properties have not been fully deciphered. In this paper, we applied the brain-wide positioning system (BPS) to image the entire brain of two Thy1-eYFP H-line male mice at 0.2 μm × 0.2 μm × 1 μm voxel resolution. A total of 103 neurons were reconstructed in layers 5 and 6 of the visual cortex with single-axon-level resolution. Based on the complete topology of neurons and the inherent positioning function of the imaging method, we classified the observed neurons into six types according to their apical dendrites and somata location: star pyramidal cells in layer 5 (L5-sp), slender-tufted pyramidal cells in layer 5 (L5-st), tufted pyramidal cells in layer 5 (L5-tt), spiny stellate-like cells in layer 6 (L6-ss), star pyramidal cells in layer 6 (L6-sp), and slender-tufted pyramidal cells in layer 6 (L6-st). By examining the axonal projection patterns of individual neurons, they can be categorized into three modes: ipsilateral circuit connection neurons, callosal projection neurons and corticofugal projection neurons. Correlating the two types of classifications, we have found that there are at least two projection modes comprised in the former defined neuron types except for L5-tt. On the other hand, each projection mode may consist of four dendritic types defined in this study. The axon projection mode only partially correlates with the apical dendrite feature. This work has demonstrated a paradigm for resolving the visual cortex through single-neuron-level quantification and has shown potential to be extended to reveal the connectome of other defined sensory and motor systems.

Morphological Neuron Classification Based on Dendritic Tree Hierarchy

The shape of a neuron can reveal interesting properties about its function. Therefore, morphological neuron characterization can contribute to a better understanding of how the brain works. However, one of the great challenges of neuroanatomy is the definition of morphological properties that can be used for categorizing neurons. This paper proposes a new methodology for neuron morphological analysis by considering different hierarchies of the dendritic tree for characterizing and categorizing neuronal cells. The methodology consists in using different strategies for decomposing the dendritic tree along its hierarchies, allowing the identification of relevant parts (possibly related to specific neuronal functions) for classification tasks. A set of more than 5000 neurons corresponding to 10 classes were examined with supervised classification algorithms based on this strategy. It was found that classification accuracies similar to those obtained by using whole neurons can be achieved by considering only parts of the neurons. Branches close to the soma were found to be particularly relevant for classification.

Large-scale Exploration of Neuronal Morphologies Using Deep Learning and Augmented Reality

Recently released large-scale neuron morphological data has greatly facilitated the research in neuroinformatics. However, the sheer volume and complexity of these data pose significant challenges for efficient and accurate neuron exploration. In this paper, we propose an effective retrieval framework to address these problems, based on frontier techniques of deep learning and binary coding. For the first time, we develop a deep learning based feature representation method for the neuron morphological data, where the 3D neurons are first projected into binary images and then learned features using an unsupervised deep neural network, i.e., stacked convolutional autoencoders (SCAEs). The deep features are subsequently fused with the hand-crafted features for more accurate representation. Considering the exhaustive search is usually very time-consuming in large-scale databases, we employ a novel binary coding method to compress feature vectors into short binary codes. Our framework is validated on a public data set including 58,000 neurons, showing promising retrieval precision and efficiency compared with state-of-the-art methods. In addition, we develop a novel neuron visualization program based on the techniques of augmented reality (AR), which can help users take a deep exploration of neuron morphologies in an interactive and immersive manner.

Mining Big Neuron Morphological Data

The advent of automatic tracing and reconstruction technology has led to a surge in the number of neurons 3D reconstruction data and consequently the neuromorphology research. However, the lack of machine-driven annotation schema to automatically detect the types of the neurons based on their morphology still hinders the development of this branch of science. Neuromorphology is important because of the interplay between the shape and functionality of neurons and the far-reaching impact on the diagnostics and therapeutics in neurological disorders. This survey paper provides a comprehensive research in the field of automatic neurons classification and presents the existing challenges, methods, tools, and future directions for automatic neuromorphology analytics. We summarize the major automatic techniques applicable in the field and propose a systematic data processing pipeline for automatic neuron classification, covering data capturing, preprocessing, analyzing, classification, and retrieval. Various techniques and algorithms in machine learning are illustrated and compared to the same dataset to facilitate ongoing research in the field.

Generative models for network neuroscience: prospects and promise

Network neuroscience is the emerging discipline concerned with investigating the complex patterns of interconnections found in neural systems, and identifying principles with which to understand them. Within this discipline, one particularly powerful approach is network generative modelling, in which wiring rules are algorithmically implemented to produce synthetic network architectures with the same properties as observed in empirical network data. Successful models can highlight the principles by which a network is organized and potentially uncover the mechanisms by which it grows and develops. Here, we review the prospects and promise of generative models for network neuroscience. We begin with a primer on network generative models, with a discussion of compressibility and predictability, and utility in intuiting mechanisms, followed by a short history on their use in network science, broadly. We then discuss generative models in practice and application, paying particular attention to the critical need for cross-validation. Next, we review generative models of biological neural networks, both at the cellular and large-scale level, and across a variety of species including Caenorhabditis elegans, Drosophila, mouse, rat, cat, macaque and human. We offer a careful treatment of a few relevant distinctions, including differences between generative models and null models, sufficiency and redundancy, inferring and claiming mechanism, and functional and structural connectivity. We close with a discussion of future directions, outlining exciting frontiers both in empirical data collection efforts as well as in method and theory development that, together, further the utility of the generative network modelling approach for network neuroscience.

Universal features of dendrites through centripetal branch ordering

Dendrites form predominantly binary trees that are exquisitely embedded in the networks of the brain. While neuronal computation is known to depend on the morphology of dendrites, their underlying topological blueprint remains unknown. Here, we used a centripetal branch ordering scheme originally developed to describe river networks—the Horton-Strahler order (SO)–to examine hierarchical relationships of branching statistics in reconstructed and model dendritic trees. We report on a number of universal topological relationships with SO that are true for all binary trees and distinguish those from SO-sorted metric measures that appear to be cell type-specific. The latter are therefore potential new candidates for categorising dendritic tree structures. Interestingly, we find a faithful correlation of branch diameters with centripetal branch orders, indicating a possible functional importance of SO for dendritic morphology and growth. Also, simulated local voltage responses to synaptic inputs are strongly correlated with SO. In summary, our study identifies important SO-dependent measures in dendritic morphology that are relevant for neural function while at the same time it describes other relationships that are universal for all dendrites.

Similarly to river beds, dendritic trees of nerve cells form elaborate networks that branch out to cover extensive areas. In the 1940s, ecologist Robert E. Horton developed an ordering system for branches in river networks that was refined in the 1950s by geoscientist Arthur N. Strahler, the Horton-Strahler order (SO). Branches at the tips start with order 1 and increase their order in a systematic way when encountering new branches on the way to the root. SO relationships have recently become popular for quantifying dendritic morphologies. Various branching statistics can be studied as a function of SO. Here we describe that topological measures such as the number of branches, the branch bifurcation ratio and the size of subtrees exhibit stereotypical relations with SO in dendritic trees independently of cell type, mirroring universal features of binary trees. Other functionally more relevant features such as mean branch lengths, local diameters and simulated voltage responses to synaptic inputs directly correlate with SO in a cell type-specific manner, indicating the importance of SO for understanding dendrite growth as well as neural computation.

Phase transitions in Pareto optimal complex networks

The organization of interactions in complex systems can be described by networks connecting different units. These graphs are useful representations of the local and global complexity of the underlying systems. The origin of their topological structure can be diverse, resulting from different mechanisms including multiplicative processes and optimization. In spatial networks or in graphs where cost constraints are at work, as it occurs in a plethora of situations from power grids to the wiring of neurons in the brain, optimization plays an important part in shaping their organization. In this paper we study network designs resulting from a Pareto optimization process, where different simultaneous constraints are the targets of selection. We analyze three variations on a problem, finding phase transitions of different kinds. Distinct phases are associated with different arrangements of the connections, but the need of drastic topological changes does not determine the presence or the nature of the phase transitions encountered. Instead, the functions under optimization do play a determinant role. This reinforces the view that phase transitions do not arise from intrinsic properties of a system alone, but from the interplay of that system with its external constraints.

Using Pareto optimality to explore the topology and dynamics of the human connectome

Graph theory has provided a key mathematical framework to analyse the architecture of human brain networks. This architecture embodies an inherently complex relationship between connection topology, the spatial arrangement of network elements, and the resulting network cost and functional performance. An exploration of these interacting factors and driving forces may reveal salient network features that are critically important for shaping and constraining the brain's topological organization and its evolvability. Several studies have pointed to an economic balance between network cost and network efficiency with networks organized in an ‘economical’ small-world favouring high communication efficiency at a low wiring cost. In this study, we define and explore a network morphospace in order to characterize different aspects of communication efficiency in human brain networks. Using a multi-objective evolutionary approach that approximates a Pareto-optimal set within the morphospace, we investigate the capacity of anatomical brain networks to evolve towards topologies that exhibit optimal information processing features while preserving network cost. This approach allows us to investigate network topologies that emerge under specific selection pressures, thus providing some insight into the selectional forces that may have shaped the network architecture of existing human brains.

A framework for analyzing the relationship between gene expression and morphological, topological, and dynamical patterns in neuronal networks

BACKGROUND

A key point in developmental biology is to understand how gene expression influences the morphological and dynamical patterns that are observed in living beings.

NEW METHOD

In this work we propose a methodology capable of addressing this problem that is based on estimating the mutual information and Pearson correlation between the intensity of gene expression and measurements of several morphological properties of the cells. A similar approach is applied in order to identify effects of gene expression over the system dynamics. Neuronal networks were artificially grown over a lattice by considering a reference model used to generate artificial neurons. The input parameters of the artificial neurons were determined according to two distinct patterns of gene expression and the dynamical response was assessed by considering the integrate-and-fire model.

RESULTS

As far as single gene dependence is concerned, we found that the interaction between the gene expression and the network topology, as well as between the former and the dynamics response, is strongly affected by the gene expression pattern. In addition, we observed a high correlation between the gene expression and some topological measurements of the neuronal network for particular patterns of gene expression.

COMPARISON WITH EXISTING METHODS

To our best understanding, there are no similar analyses to compare with.

CONCLUSIONS

A proper understanding of gene expression influence requires jointly studying the morphology, topology, and dynamics of neurons. The proposed framework represents a first step towards predicting gene expression patterns from morphology and connectivity.

Genotype-specific effects of Mecp2 loss-of-function on morphology of Layer V pyramidal neurons in heterozygous female Rett syndrome model mice

Rett syndrome (RTT) is a progressive neurological disorder primarily caused by mutations in the X-linked gene methyl-CpG-binding protein 2 (MECP2). The heterozygous female brain consists of mosaic of neurons containing both wild-type MeCP2 (MeCP2+) and mutant MeCP2 (MeCP2-). Three-dimensional morphological analysis was performed on individually genotyped layer V pyramidal neurons in the primary motor cortex of heterozygous (Mecp2(+/-) ) and wild-type (Mecp2(+/+) ) female mice ( > 6 mo.) from the Mecp2(tm1.1Jae) line. Comparing basal dendrite morphology, soma and nuclear size of MeCP2+ to MeCP2- neurons reveals a significant cell autonomous, genotype specific effect of Mecp2. MeCP2- neurons have 15% less total basal dendritic length, predominantly in the region 70-130 μm from the cell body and on average three fewer branch points, specifically loss in the second and third branch orders. Soma and nuclear areas of neurons of mice were analyzed across a range of ages (5-21 mo.) and X-chromosome inactivation (XCI) ratios (12-56%). On average, MeCP2- somata and nuclei were 15 and 13% smaller than MeCP2+ neurons respectively. In most respects branching morphology of neurons in wild-type brains (MeCP2 WT) was not distinguishable from MeCP2+ but somata and nuclei of MeCP2 WT neurons were larger than those of MeCP2+ neurons. These data reveal cell autonomous effects of Mecp2 mutation on dendritic morphology, but also suggest non-cell autonomous effects with respect to cell size. MeCP2+ and MeCP2- neuron sizes were not correlated with age, but were correlated with XCI ratio. Unexpectedly the MeCP2- neurons were smallest in brains where the XCI ratio was highly skewed toward MeCP2+, i.e., wild-type. This raises the possibility of cell non-autonomous effects that act through mechanisms other than globally secreted factors; perhaps competition for synaptic connections influences cell size and morphology in the genotypically mosaic brain of RTT model mice.

Statistical analysis and data mining of digital reconstructions of dendritic morphologies

Neuronal morphology is diverse among animal species, developmental stages, brain regions, and cell types. The geometry of individual neurons also varies substantially even within the same cell class. Moreover, specific histological, imaging, and reconstruction methodologies can differentially affect morphometric measures. The quantitative characterization of neuronal arbors is necessary for in-depth understanding of the structure-function relationship in nervous systems. The large collection of community-contributed digitally reconstructed neurons available at NeuroMorpho.Org constitutes a “big data” research opportunity for neuroscience discovery beyond the approaches typically pursued in single laboratories. To illustrate these potential and related challenges, we present a database-wide statistical analysis of dendritic arbors enabling the quantification of major morphological similarities and differences across broadly adopted metadata categories. Furthermore, we adopt a complementary unsupervised approach based on clustering and dimensionality reduction to identify the main morphological parameters leading to the most statistically informative structural classification. We find that specific combinations of measures related to branching density, overall size, tortuosity, bifurcation angles, arbor flatness, and topological asymmetry can capture anatomically and functionally relevant features of dendritic trees. The reported results only represent a small fraction of the relationships available for data exploration and hypothesis testing enabled by sharing of digital morphological reconstructions.

Big data from small data: data-sharing in the 'long tail' of neuroscience

The launch of the US BRAIN and European Human Brain Projects coincides with growing international efforts toward transparency and increased access to publicly funded research in the neurosciences. The need for data-sharing standards and neuroinformatics infrastructure is more pressing than ever. However, 'big science' efforts are not the only drivers of data-sharing needs, as neuroscientists across the full spectrum of research grapple with the overwhelming volume of data being generated daily and a scientific environment that is increasingly focused on collaboration. In this commentary, we consider the issue of sharing of the richly diverse and heterogeneous small data sets produced by individual neuroscientists, so-called long-tail data. We consider the utility of these data, the diversity of repositories and options available for sharing such data, and emerging best practices. We provide use cases in which aggregating and mining diverse long-tail data convert numerous small data sources into big data for improved knowledge about neuroscience-related disorders.

Quantitative investigations of axonal and dendritic arbors: development, structure, function, and pathology

The branching structures of neurons are a long-standing focus of neuroscience. Axonal and dendritic morphology affect synaptic signaling, integration, and connectivity, and their diversity reflects the computational specialization of neural circuits. Altered neuronal morphology accompanies functional changes during development, experience, aging, and disease. Technological improvements continuously accelerate high-throughput tissue processing, image acquisition, and morphological reconstruction. Digital reconstructions of neuronal morphologies allow for complex quantitative analyses that are unattainable from raw images or two-dimensional tracings. Furthermore, digitized morphologies enable computational modeling of biophysically realistic neuronal dynamics. Additionally, reconstructions generated to address specific scientific questions have the potential for continued investigations beyond the original reason for their acquisition. Facilitating multiple reuse are repositories like NeuroMorpho.Org, which ease the sharing of reconstructions. Here, we review selected scientific literature reporting the reconstruction of axonal or dendritic morphology with diverse goals including establishment of neuronal identity, examination of physiological properties, and quantification of developmental or pathological changes. These reconstructions, deposited in NeuroMorpho.Org, have since been used by other investigators in additional research, of which we highlight representative examples. This cycle of data generation, analysis, sharing, and reuse reveals the vast potential of digital reconstructions in quantitative investigations of neuronal morphology.

A software tool for the analysis of neuronal morphology data

Anatomy plays a fundamental role in supporting and shaping nervous system activity. The remarkable progress of computer processing power within the last two decades has enabled the generation of electronic databases of complete three-dimensional (3D) dendritic and axonal morphology for neuroanatomical studies. Several laboratories are freely posting their reconstructions online after result publication v.gr. NeuroMorpho.Org (Nat Rev Neurosci7:318-324, 2006). These neuroanatomical archives represent a crucial resource to explore the relationship between structure and function in the brain (Front Neurosci6:49, 2012). However, such 'Cartesian' descriptions bear little intuitive information for neuroscientists. Here, we developed a simple prototype of a MATLAB-based software tool to quantitatively describe the 3D neuronal structures from public repositories. The program imports neuronal reconstructions and quantifies statistical distributions of basic morphological parameters such as branch length, tortuosity, branch's genealogy and bifurcation angles. Using these morphological distributions, our algorithm can generate a set of virtual neurons readily usable for network simulations.

Neuronal morphology goes digital: a research hub for cellular and system neuroscience

The importance of neuronal morphology in brain function has been recognized for over a century. The broad applicability of "digital reconstructions" of neuron morphology across neuroscience subdisciplines has stimulated the rapid development of numerous synergistic tools for data acquisition, anatomical analysis, three-dimensional rendering, electrophysiological simulation, growth models, and data sharing. Here we discuss the processes of histological labeling, microscopic imaging, and semiautomated tracing. Moreover, we provide an annotated compilation of currently available resources in this rich research "ecosystem" as a central reference for experimental and computational neuroscience.