- Abstract: To understand functions of a single neuron, such as propagation and generation of synaptic or action potentials, a detailed description of the kinetics and distribution of the underlying ionic conductances is essential. In voltage-clamp experiments, incomplete space clamp distorts the recorded currents, rendering accurate analysis impossible. Here, we present a simple numerical algorithm that corrects such distortions. The method performs a stepwise approximation of the conductance density... Show More
Conference Paper: Supervised Learning of Image Restoration with Convolutional NetworksAbstract: Convolutional networks have achieved a great deal of success in high-level vision problems such as object recognition. Here we show that they can also be used as a general method for low-level image processing. As an example of our approach, convolutional networks are trained using gradient learning to solve the problem of restoring noisy or degraded images. For our training data, we have used electron microscopic images of neural circuitry with ground truth restorations provided by human... Show More
- Abstract: Voltage-gated potassium channels effectively regulate dendritic excitability in neurones. It has been suggested that in the distal apical dendrite of layer 5B (L5B) neocortical pyramidal neurones, K+ conductances participate in active dendritic synaptic integration and control regenerative dendritic potentials. The ionic mechanism for triggering these regenerative potentials has yet to be elucidated. Here we used two-electrode voltage clamp (TEVC) to quantitatively record K+ conductance... Show More
- Abstract: The characterization of individual neurons by Golgi and Cajal has been the basis of neuroanatomy for a century. A new challenge is to anatomically describe, at cellular resolution, complete local circuits that can drive behavior. In this essay, we review the possibilities to obtain a model cortical column by using in vitro and in vivo pair recordings, followed by anatomical reconstructions of the projecting and target cells. These pairs establish connection modules that eventually may be... Show More
- Abstract: Layer 5 (L5) of somatosensory cortex is a major gateway for projections to intra- and subcortical brain regions. This layer is further divided into 5A and 5B characterized by relatively separate afferent and efferent connections. Little is known about the organization of connections within L5A of neocortical columns. We therefore used paired recordings to probe the anatomy and physiology of monosynaptic connections between L5A pyramidal neurons within the barrel columns of somatosensory... Show More
- Abstract: Three-dimensional electron-microscopic image stacks with almost isotropic resolution allow, for the first time, to determine the complete connection matrix of parts of the brain. In spite of major advances in staining, correct segmentation of these stacks remains challenging, because very few local mistakes can lead to severe global errors. We propose a hierarchical segmentation procedure based on statistical learning and topology-preserving grouping. Edge probability maps are computed by a... Show More
- Abstract: Neuroscience produces a vast amount of data from an enormous diversity of neurons. A neuronal classification system is essential to organize such data and the knowledge that is derived from them. Classification depends on the unequivocal identification of the features that distinguish one type of neuron from another. The problems inherent in this are particularly acute when studying cortical interneurons. To tackle this, we convened a representative group of researchers to agree on a set of... Show More
- Abstract: Interneurons in layers 2/3 are excited by pyramidal cells within the same layer (Reyes et al., 1998; Gupta et al., 2000), but little is known about translaminar innervation of these interneurons by spiny neurons in the main cortical input layer 4 (L4). Here, we investigated (1) how efficiently L4 spiny neurons excite L2/3 interneurons via monosynaptic connections, (2) whether glutamate release from axon terminals of L4 spiny neurons depends on the identity of the postsynaptic interneuron,... Show More
- Abstract: Interneurons in layer 2/3 (L2/3) of the somatosensory cortex show 4 types of axonal projection patterns with reference to the laminae and borders of columns in rat barrel cortex (Helmstaedter et al. 2008a). Here, we analyzed the dendritic geometry and electrical excitability of these interneurons. First, dendritic polarity, measured based on the insertion points of primary dendrites on the soma surface, yielded a continuous one-dimensional measure without a clustering of dendritic polarity... Show More
- Abstract: For a detailed description of the circuitry of cortical columns at the level of single neurons, it is essential to define the identities of the cell types that constitute these columns. For interneurons (INs), we described 4 "types of axonal projection patterns" in layer 2/3 (L2/3) with reference to the outlines of a cortical column (Helmstaedter et al. 2008a). In addition we quantified the dendritic geometry and electrical excitability of 3 types of the L2/3 INs: "local," "lateral," and... Show More
- Abstract: In the neocortex, inhibition by gamma-aminobutyric acidergic (GABAergic) interneurons is essential for shaping cortical maps, which represent sensory signals. For a detailed understanding of the stream of excitation evoked, for example, by a sensory stimulus, interneurons must be identified with reference to their impact on excitatory neurons located in different laminae of the same (home) and surround columns. We analyzed the axonal projection of layer 2/3 (L2/3) interneurons with reference... Show More
- Abstract: Recent technological developments have renewed the interest in large-scale neural circuit reconstruction. To resolve the structure of entire circuits, thousands of neurons must be reconstructed and their synapses identified. Reconstruction techniques at the light microscopic level are capable of following sparsely labeled neurites over long distances, but fail with densely labeled neuropil. Electron microscopy provides the resolution required to resolve densely stained neuropil, but is... Show More
- Abstract: Many image segmentation algorithms first generate an affinity graph and then partition it. We present a machine learning approach to computing an affinity graph using a convolutional network (CN) trained using ground truth provided by human experts. The CN affinity graph can be paired with any standard partitioning algorithm and improves segmentation accuracy significantly compared to standard hand-designed affinity functions. We apply our algorithm to the challenging 3D segmentation... Show More
- Abstract: Images can be segmented by first using a classifier to predict an affinity graph that reflects the degree to which image pixels must be grouped together and then partitioning the graph to yield a segmentation. Machine learning has been applied to the affinity classifier to produce affinity graphs that are good in the sense of minimizing edge misclassification rates. However, this error measure is only indirectly related to the quality of segmentations produced by ultimately partitioning the... Show More
- Abstract: The axonal projections of nerve cells have been used to infer synaptic connectivity ever since the drawings of Ramon y Cajal more than a hundred years ago. Here we review the assumptions behind these studies and report how axonal projections of thalamic and cortical neurons can be used to anatomically define cortical columns as innervation volumes in rat barrel cortex. We then apply this analysis to cortical interneurons and illustrate that it is the axonal projections of interneurons which... Show More
Data: Supplementary Data
Data: Supplementary Data
- Abstract: Recent studies have shown that machine learning can improve the accuracy of detecting object boundaries in images. In the standard approach, a boundary detector is trained by minimizing its pixel-level disagreement with human boundary tracings. This naive metric is problematic because it is overly sensitive to boundary locations. This problem is solved by metrics provided with the Berkeley Segmentation Dataset, but these can be insensitive to topological differences, such as gaps in... Show More
Publications citing this author (3251)
[Show abstract] [Hide abstract] ABSTRACT: State-of-the-art image segmentation algorithms generally consist of at least two successive and distinct computations: a boundary detection process that uses local image information to classify image locations as boundaries between objects, followed by a pixel grouping step such as watershed or connected components that clusters pixels into segments. Prior work has varied the complexity and approach employed in these two steps, including the incorporation of multi-layer neural networks to perform boundary prediction, and the use of global optimizations during pixel clustering. We propose a unified and end-to-end trainable machine learning approach, flood-filling networks, in which a recurrent 3d convolutional network directly produces individual segments from a raw image. The proposed approach robustly segments images with an unknown and variable number of objects as well as highly variable object sizes. We demonstrate the approach on a challenging 3d image segmentation task, connectomic reconstruction from volume electron microscopy data, on which flood-filling neural networks substantially improve accuracy over other state-of-the-art methods. The proposed approach can replace complex multi-step segmentation pipelines with a single neural network that is learned end-to-end.
- Although segmentation quality is often evaluated with respect to ground truth pixel-wise contours or object masks , creating such ground truth for large-scale connectomic datasets that span billions or trillions of voxels is highly laborious. A more efficient way to generate ground truth representations of large-scale neuron topology is to " skeletonize " neurons into a collection of points that typically constitute an undirected tree .
[Show abstract] [Hide abstract] ABSTRACT: Synaptic connectivity between neurons is naturally constrained by the anatomical overlap of neuronal arbors, the space on the axon available for synapses, and by physiological mechanisms that form synapses at a subset of potential synapse locations. What is not known is how these constraints impact emergent connectivity in a circuit with diverse morphologies. We investigated the role of morphological diversity within and across neuronal types on emergent connectivity in a model of neocortical microcircuitry. We found that the average overlap between the dendritic and axonal arbors of different types of neurons determines neuron-type specific patterns of distance-dependent connectivity, severely constraining the space of possible connectomes. However, higher order connectivity motifs depend on the diverse branching patterns of individual arbors of neurons belonging to the same type. Morphological diversity across neuronal types, therefore, imposes a specific structure on first order connectivity, and morphological diversity within neuronal types imposes a higher order structure of connectivity. We estimate that the morphological constraints resulting from diversity within and across neuron types together lead to a 10-fold reduction of the entropy of possible connectivity configurations, revealing an upper bound on the space explored by structural plasticity.
- Structure in the connectivity of neural microcircuitry extends beyond laminarity, with different neuron types within a layer preferentially sending synaptic output to and receiving input from a subset of other neuron types in the same and other layers (Silberberg and Markram 2007;Helmstaedter et al. 2008;Brown and Hestrin 2009). This pattern is frequently explained in terms of specific molecular targeting mechanisms and experience-driven plasticity (Somogyi 1977;Somogyi et al. 1998;Hofer et al. 2009;Holtmaat and Svoboda 2009).
[Show abstract] [Hide abstract] ABSTRACT: Present neurobiological concepts regarding superior cognitive functions are based on synaptic neurotransmission and neuronal plasticity. However, the diversity and complexity of neuro-cortical connections, circuits, maps and their relationships with memory, learning and other superior cognitive functions are not fully explained by the present neurobiological paradigms. Recent discoveries concerning the ability of neuronal cells to perceive and process very-weak electromagnetic information suggest a possible role of bio-photons generated as consequence of neuronal and astroglial metabolisms in a wide diversity of cognitive representations. Moreover, the finding that human brain has magnetite nanoparticles opens new possibilities about the role of these nano-crystals in information processing and memory. In the present chapter, a previously advanced model (Neuron-Astroglial Communication in Short-Term Memory: Bio-Electric, Bio-Magnetic and Bio-Photonic Signals?) is developed. Based in the ability of neurons and astrocytes to generate bio-photons as consequence of their metabolic activities, I propose the generation of innumerable bio-phonic-mediated nano-holograms, which are produced and modulated by magnetite nano-crystals associated to neuronal and astroglial membranes in the cerebral neocortex. Specifically, it is suggested that bio-photons generated by neuronal and astroglial cells may produce multichannel holographic pictures through their interaction with single domain and/or superparamagnetic magnetite nanoparticles, explaining retrieval of short-term and long-term memories as well as other neuro- cognitive representations such as the ―images‖ generated in dreams. Bio-chemic, bio- electric, bio-magnetic and bio-photonic activities in the cerebral cortex are not independent bio-physical phenomena, suggesting that interactions among these signals may contribute to information exchange and processing in the neocortex. This hypothesis proposes that the interactions among bio-chemic, bio-electric, bio-magnetic and bio- photonic activities in neurons and astrocytes in the human cerebral neocortex are not epiphenomena of the cerebral activity but they play important roles in cognitive functions, providing new perspectives for better understand complex cognitive functions.
- Despite it has been observed a considerable diversity of cell types, laminar organization, functional architecture and connectivity, cerebral neocortex seems to be composed by innumerable neuronal and glial cells organized fundamentally in two closely interrelated 3-D structures (Figure 1). One formed by the neurocortex, organized in neuronal layers and columns (Buxhoeveden and Casanova, 2002; De Felipe et al., 2002; Helmstaedter et al., 2007; Hubel and Wiesel, 1969; Livingstone et al., 1995; Livingstone and Hubel, 1984; Lorente De Nó, 1949; MarinPadilla, 1992; Mountcastle, 1957; Ramon y Cajal, 1897); the other by the astroglial syncytium or gliocortex (Mugniani, 1986). This fundamental binary organization of the neocortex has been proposed as the central concept for understand information processing, memory and other superior cognitive representations (Banacclocha, 2004Banacclocha, , 2005Banacclocha, , 2007Banacclocha, , 2010Banacclocha, , 2011Banacclocha, , 2012) and it is now considered essential for the present model.
[Show abstract] [Hide abstract] ABSTRACT: Cortical inhibitory neurons exhibit remarkable diversity in their morphology, connectivity, and synaptic properties. Here, we review the function of somatostatin-expressing (SOM) inhibitory interneurons, focusing largely on sensory cortex. SOM neurons also comprise a number of subpopulations that can be distinguished by their morphology, input and output connectivity, laminar location, firing properties, and expression of molecular markers. Several of these classes of SOM neurons show unique dynamics and characteristics, such as facilitating synapses, specific axonal projections, intralaminar input, and top-down modulation, which suggest possible computational roles. SOM cells can be differentially modulated by behavioral state depending on their class, sensory system, and behavioral paradigm. The functional effects of such modulation have been studied with optogenetic manipulation of SOM cells, which produces effects on learning and memory, task performance, and the integration of cortical activity. Different classes of SOM cells participate in distinct disinhibitory circuits with different inhibitory partners and in different cortical layers. Through these disinhibitory circuits, SOM cells help encode the behavioral relevance of sensory stimuli by regulating the activity of cortical neurons based on subcortical and intracortical modulatory input. Associative learning leads to long-term changes in the strength of connectivity of SOM cells with other neurons, often influencing the strength of inhibitory input they receive. Thus despite their heterogeneity and variability across cortical areas, current evidence shows that SOM neurons perform unique neural computations, forming not only distinct molecular but also functional subclasses of cortical inhibitory interneurons.
- Based on histochemical markers, we can divide cortical inhibitory cells into three non-overlapping categories in mice: those that express parvalbumin (PV), somatostatin (SOM), or vasointestinal peptide (VIP). These categories vary across species; in rats, for example, PV, SOM, and calretinin (CR) cells form non-overlapping categories (Gonchar and Burkhalter, 1997; Kawaguchi and Kubota, 1997), whereas mice show overlapping expression of SOM and CR (Freund and Buzsáki, 1996; Somogyi and Klausberger, 2005; Ascoli et al., 2008; Fishell and Rudy, 2011; Rudy et al., 2011). While these 3 major classes don't account for all inhibitory interneurons (a small number of interneurons express other less common markers), these 3 major classes do account for the vast majority (80–90%) of all inhibitory cells (Gonchar and Burkhalter, 1997; Rudy et al., 2011; Pfeffer et al., 2013).
[Show abstract] [Hide abstract] ABSTRACT: The remarkable development of deep learning in medicine and healthcare domain presents obvious privacy issues, when deep neural networks are built on users' personal and highly sensitive data, e.g., clinical records, user profiles, biomedical images, etc. However, only a few scientific studies on preserving privacy in deep learning have been conducted. In this paper, we focus on developing a private convolutional deep belief network (pCDBN), which essentially is a convolutional deep belief network (CDBN) under differential privacy. Our main idea of enforcing $\epsilon$-differential privacy is to leverage the functional mechanism to perturb the energy-based objective functions of traditional CDBNs, rather than their results. One key contribution of this work is that we propose the use of Chebyshev expansion to derive the approximate polynomial representation of objective functions. Our theoretical analysis shows that we can further derive the sensitivity and error bounds of the approximate polynomial representation. As a result, preserving differential privacy in CDBNs is feasible. We applied our model in a health social network, i.e., YesiWell data, and in a handwriting digit dataset, i.e., MNIST data, for human behavior prediction, human behavior classification, and handwriting digit recognition tasks. Theoretical analysis and rigorous experimental evaluations show that the pCDBN is highly effective. It significantly outperforms existing solutions.
- That makes it easier to extract useful information when building classifiers and predictors (LeCun et al. 2015). Deep learning has applications in a number of healthcare areas, e.g., phenotype extraction and health risk prediction (Cheng et al. 2016), prediction of the development of various diseases including schizophrenia, a variety of cancers, diabetes, heart failure, etc. (Choi et al. 2016;Li et al. 2015;Miotto et al. 2016;Roumia and Steinhubl 2014;Wu et al. 2010), prediction of risk of readmission (Wu et al. 2010), Alzheimer's diagnosis (Liu et al. 2014;Ortiz et al. 2016), risk prediction for chronic kidney disease progression (Perotte et al. 2015), physical activity prediction (Phan et al. 2015aPhan et al. , b, 2016a, c), feature learning from fMRI data (Plis et al. 2014), diagnosis code assignment (Gottlieb et al. 2013;Perotte et al. 2014), reconstruction of brain circuits (Helmstaedter et al. 2013), prediction of the activity of potential drug molecules (Ma et al. 2015), the effects of mutations in non-coding DNA on gene expressions (Leung et al. 2014;Xiong et al. 2015), and many more. The development of deep learning in the domain of medicine and healthcare presents obvious privacy issues, when deep neural networks are built based on patients' personal and highly sensitive data, e.g., clinical records, user profiles, biomedical images, etc.
[Show abstract] [Hide abstract] ABSTRACT: An essential step toward understanding neural circuits is linking their structure and their dynamics. In general, this relationship can be almost arbitrarily complex. Recent theoretical work has, however, begun to identify some broad principles underlying collective spiking activity in neural circuits. The first is that local features of network connectivity can be surprisingly effective in predicting global statistics of activity across a network. The second is that, for the important case of large networks with excitatory-inhibitory balance, correlated spiking persists or vanishes depending on the spatial scales of recurrent and feedforward connectivity. We close by showing how these ideas, together with plasticity rules, can help to close the loop between network structure and activity statistics.
- At first, the path to understanding these mechanisms seems extremely complicated. Electron microscopy (EM) and allied reconstruction methods promise connectomes among thousands of nearby cells, tabulating an enormous amount of data . This begs the question of what statistics of connectivity matter most – and least – in driving the important activity patterns of neural populations.
[Show abstract] [Hide abstract] ABSTRACT: EEG is the oldest tool for studying human cognitive function, but it is blamed by many to be useless because its poor spatial resolution despite it excellent temporal discrimination. Such comments arise from a reductionist point of view about the cerebral function. If the brain is assumed to be a distributed processing system recruiting different types of neurons and astrocytes widely locate in many distinct cerebral neighborhoods then EEG temporal resolution and the many different tools available for its analysis, turn it the tool of choice to investigate human cognition. The present paper follows this point of view. Here, to better understand the different types of information encoded in the recorded cortical electrical activity a clear model of the cortical structure and function of the different cortical column layers is presented and discussed. Taking this model into consideration, different available techniques for EEG analysis are presented and discussed under the assumption that tool combination is a necessity to provide a full comprehension of dynamics of the cortical activity supporting human cognition. Data on language understanding published elsewhere are used to illustrate how to use this combined and complex EEG analysis to disclose important details of cognitive cerebral dynamics, which reveal that cognitive neural circuits are scale free networks supporting entrainment of a large number of cortical column assemblies distributed all over the cortex. Reasoning is then assumed to result from such well orchestrated large scale entrainment. how to cite this paper A. F. Rocha Toward a comprehensive understanding of EEG and its analyses Cognitive Neurosciences ejournal, (2017), June 14 https://papers.ssrn.com/abstract=2975753
- Astrocytes make contacts to both pre-synaptic terminals and postsynaptic dendrites and are also influential upon neuronal activity (e.g.,Pereira and Furlan, 2010). Oscillatory activity at the different layers (Figure 2) may be dependent on their membrane properties or due to local neural circuits involving cells of the same or neighboring columns (Contreras, D andLlinás, 2001;Moritz et al, 2009;Shuzo, and Harris , 2009). Input activity from thalamus another important influence over columnar electrical activity, mostly the oscillatory activity originating on the reticular thalamic neurons (McCormick, et al, 2015;Olejniczak, 2006; Rocha, Pereira Jr. and Coutinho2 2001 and Steriade, McCormic and,Sejnowski , 1993).
[Show abstract] [Hide abstract] ABSTRACT: Perceptual decisions involve distributed cortical activity. Does information flow sequentially from one cortical area to another, or do networks of interconnected areas contribute at the same time? Here we delineate when and how activity in specific areas drives a whisker-based decision in mice. A short-term memory component temporally separated tactile "sensation" and "action" (licking). Using optogenetic inhibition (spatial resolution, 2 mm; temporal resolution, 100 ms), we surveyed the neocortex for regions driving behavior during specific behavioral epochs. Barrel cortex was critical for sensation. During the short-term memory, unilateral inhibition of anterior lateral motor cortex biased responses to the ipsilateral side. Consistently, barrel cortex showed stimulus-specific activity during sensation, whereas motor cortex showed choice-specific preparatory activity and movement-related activity, consistent with roles in motor planning and movement. These results suggest serial information flow from sensory to motor areas during perceptual decision making.
- To probe the role of specific brain areas, we inactivated small volumes of cortical tissue by photostimulating channelrhodop- sin-2 (ChR2) in GABAergic interneurons (VGAT-ChR2-EYFP) (Zhao et al., 2011) (Figure S2). Since GABAergic interneurons have dense local axonal arbors (Helmstaedter et al., 2009), this approach is expected to produce potent and local inhibition (Figure 2A). We characterized ''photoinhibition'' in awake mice (Figure S3; see Experimental Procedures).
[Show abstract] [Hide abstract] ABSTRACT: For the past decade, convolutional networks have been used for 3D reconstruction of neurons from electron microscopic (EM) brain images. Recent years have seen great improvements in accuracy, as evidenced by submissions to the SNEMI3D benchmark challenge. Here we report the first submission to surpass the estimate of human accuracy provided by the SNEMI3D leaderboard. A variant of 3D U-Net is trained on a primary task of predicting affinities between nearest neighbor voxels, and an auxiliary task of predicting long-range affinities. The training data is augmented by simulated image defects. The nearest neighbor affinities are used to create an oversegmentation, and then supervoxels are greedily agglomerated based on mean affinity. The resulting SNEMI3D score exceeds the estimate of human accuracy by a large margin. While one should be cautious about extrapolating from the SNEMI3D benchmark to real-world accuracy of large-scale neural circuit reconstruction, our result inspires optimism that the goal of full automation may be realizable in the future.
- Turaga et al.trained convolutional networks to transform an input EM image stack into an output affinity graph, which is subsequently partitioned to produce a segmentation. They included only edges between nearest neighbor voxels in the affinity graph.
[Show abstract] [Hide abstract] ABSTRACT: Learned boundary maps are known to outperform hand- crafted ones as a basis for the watershed algorithm. We show, for the first time, how to train watershed computation jointly with boundary map prediction. The estimator for the merging priorities is cast as a neural network that is con- volutional (over space) and recurrent (over iterations). The latter allows learning of complex shape priors. The method gives the best known seeded segmentation results on the CREMI segmentation challenge.
- Such a network was successfully used as a basis for watershed segmentation in. The MALIS algorithmcomputes shortest paths between pairs of nodes and applies a correction to the highest edge along paths affected by false splits or mergers. This is similar to our training, but we apply corrections to root error edges as defined below.
[Show abstract] [Hide abstract] ABSTRACT: Images projected onto the retina of an animal eye are rarely still. Instead, they usually contain motion signals originating either from moving objects or from retinal slip caused by self-motion. Accordingly, motion signals tell the animal in which direction a predator, prey, or the animal itself is moving. At the neural level, visual motion detection has been proposed to extract directional information by a delay-and-compare mechanism, representing a classic example of neural computation. Neurons responding selectively to motion in one but not in the other direction have been identified in many systems, most prominently in the mammalian retina and the fly optic lobe. Technological advances have now allowed researchers to characterize these neurons' upstream circuits in exquisite detail. Focusing on these upstream circuits, we review and compare recent progress in understanding the mechanisms that generate direction selectivity in the early visual system of mammals and flies. Expected final online publication date for the Annual Review of Neuroscience Volume 40 is July 8, 2017. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.
- Thus, quadrants of SAC dendrites behave like local motion detectors. Second,Briggman et al. (2011)used calcium imaging and electron microscopy (EM) to reveal that DSGCs receive inhibition predominantly from SAC dendrites pointing in their null direction. For example, a nasally preferring DSGC receives more inputs from SAC dendrites pointing in the temporal direction.
[Show abstract] [Hide abstract] ABSTRACT: To study the spatiotemporal dynamics of neural activity in a cortical population, we implanted a 10x10 microelectrode array in the vibrissal cortex of urethane-anesthetized rats. We recorded spontaneous neuronal activity as well as activity evoked in response to sustained and brief sensory stimulation. To quantify the temporal dynamics of activity, we computed the probability distribution function (PDF) of spiking on one electrode given the observation of a spike on another. The spike-triggered PDFs quantified the strength, temporal delay, and temporal precision of correlated activity across electrodes. Nearby cells showed higher levels of correlation at short delays, whereas distant cells showed lower levels of correlation, which tended to occur at longer delays. We found that functional space built based on the strength of pairwise correlations predicted the anatomical arrangement of electrodes. Moreover, the correlation profile of electrode pairs during spontaneous activity predicted the “signal” and “noise” correlations during sensory stimulation. Finally, mutual information analyses revealed that neurons with stronger correlations to the network during spontaneous activity, conveyed higher information about the sensory stimuli in their evoked response. Given the 400-µm-distance between adjacent electrodes, our functional quantifications unravel the spatiotemporal dynamics of activity among nearby and distant cortical columns.
- We recorded neural population activity across a 3.6 × 3.6 mm 2 area of the rat cortex. Given the complex three-dimensional architecture of cortical columns (Egger et al., 2012) and the cortical curvature, the recordings could not be attributed to specific columns, layers, and cell types. Therefore, here we have assumed an oversimplified model of the cortex whereby all analyses focused on the spatial distance across electrode tips without including the three-dimensional structure of the cortex and the differences across neurons.
[Show abstract] [Hide abstract] ABSTRACT: Background Imaging large volumes such as entire cells or small model organisms at nanoscale resolution seemed an unrealistic, rather tedious task so far. Now, technical advances have lead to several electron microscopy (EM) large volume imaging techniques. One is array tomography, where ribbons of ultrathin serial sections are deposited on solid substrates like silicon wafers or glass coverslips. ResultsTo ensure reliable retrieval of multiple ribbons from the boat of a diamond knife we introduce a substrate holder with 7 axes of translation or rotation specifically designed for that purpose. With this device we are able to deposit hundreds of sections in an ordered way in an area of 22 × 22 mm, the size of a coverslip. Imaging such arrays in a standard wide field fluorescence microscope produces reconstructions with 200 nm lateral resolution and 100 nm (the section thickness) resolution in z.By hierarchical imaging cascades in the scanning electron microscope (SEM), using a new software platform, we can address volumes from single cells to complete organs. In our first example, a cell population isolated from zebrafish spleen, we characterize different cell types according to their organelle inventory by segmenting 3D reconstructions of complete cells imaged with nanoscale resolution. In addition, by screening large numbers of cells at decreased resolution we can define the percentage at which different cell types are present in our preparation. With the second example, the root tip of cress, we illustrate how combining information from intermediate resolution data with high resolution data from selected regions of interest can drastically reduce the amount of data that has to be recorded. By imaging only the interesting parts of a sample considerably less data need to be stored, handled and eventually analysed. Conclusions Our custom-designed substrate holder allows reproducible generation of section libraries, which can then be imaged in a hierarchical way. We demonstrate, that EM volume data at different levels of resolution can yield comprehensive information, including statistics, morphology and organization of cells and tissue. We predict, that hierarchical imaging will be a first step in tackling the big data issue inevitably connected with volume EM.
- The method was originally introduced for multiplexing immuno-staining by repeated stripping and relabelling of the section arrays in order to map synaptic connections in brain [6, 7]. In the neurosciences field, that pioneered all volume EM techniques (reviewed in ), variations of the original method are quite common, also extending it to SEM imaging (reviewed in [2, 9]). However, applications in cell and developmental or even general biology are rather scarce up to now  .
[Show abstract] [Hide abstract] ABSTRACT: A detailed inventory of the constituent pieces in cerebral cortex is considered essential to understand the principles underlying cortical signal processing. Specifically, the search for pyramidal neuron subtypes is partly motivated by the hypothesis that a subtype-specific division of labor could create a rich substrate for computation. On the other hand, the extreme integration of individual neurons into the collective cortical circuit promotes the hypothesis that cellular individuality represents a smaller computational role within the context of the larger network. These competing hypotheses raise the important question to what extent the computational function of a neuron is determined by its individual type or by its circuit connections. We created electrophysiological profiles from pyramidal neurons within the sole cellular layer of turtle visual cortex by measuring responses to current injection using whole-cell recordings. A blind clustering algorithm applied to these data revealed the presence of two principle types of pyramidal neurons. Brief diffuse light flashes triggered membrane potential fluctuations in those same cortical neurons. The apparently network driven variability of the visual responses concealed the existence of subtypes. In conclusion, our results support the notion that the importance of diverse intrinsic physiological properties is minimized when neurons are embedded in a synaptic recurrent network.
- One strategy is to reduce the subjectivity inherent to the choice of threshold by adding cells of different types (e.g. pyramidal neurons in an interneuronal classification study) to the data base [43, 49, 55, 63] or by overlaying morphology on purely electrophysiological clusters [45, 46, 64]. Another strategy is to elevate the confidence levels behind clustering by analyzing the clustering results for different threshold levels and comparing relative " accuracy " measurements across those schemes.
[Show abstract] [Hide abstract] ABSTRACT: Organism-level systems biology aims to identify, analyze, control and design cellular circuits in organisms. Many experimental and computational approaches have been developed over the years to allow us to conduct these studies. Some of the most powerful methods are based on using optical imaging in combination with fluorescent labeling, and for those one of the long-standing stumbling blocks has been tissue opacity. Recently, the solutions to this problem have started to emerge based on whole-body and whole-organ clearing techniques that employ innovative tissue-clearing chemistry. Here, we review these advancements and discuss how combining new clearing techniques with high-performing fluorescent proteins or small molecule tags, rapid volume imaging and efficient image informatics is resulting in comprehensive and quantitative organ-wide, single-cell resolution experimental data. These technologies are starting to yield information on connectivity and dynamics in cellular circuits at unprecedented resolution, and bring us closer to system-level understanding of physiology and diseases of complex mammalian systems.
- An example of the analysis pipeline is shown in Figure 5 . Procedures following this general pipeline have been proposed and applied in several studies of multiple mammalian, fish, or fly datasets of whole-organ scale (Ahrens et al., 2013; Amat et al., 2015; Helmstaedter and Mitra, 2012; Menegas et al., 2015; Portugues et al., 2014; Susaki et al., 2014 Susaki et al., , 2015). Such pipelines can be partly constructed by available software.