A Comparison Between Single- and Multi-Scale Approaches for Classification of Histopathology Images.

Whole slide images (WSIs) are digitized histopathology images. WSIs are stored in a pyramidal data structure that contains the same images at multiple magnification levels. In digital pathology, most algorithmic approaches to analyze WSIs use a single magnification level. However, images at different magnification levels may reveal relevant and distinct properties in the image, such as global context or detailed spatial arrangement. Given their high resolution, WSIs cannot be processed as a whole and are broken down into smaller pieces called tiles. Then, a prediction at the tile-level is made for each tile in the larger image. As many classification problems require a prediction at a slide-level, there exist common strategies to integrate the tile-level insights into a slide-level prediction. We explore two approaches to tackle this problem, namely a multiple instance learning framework and a representation learning algorithm (the so-called "barcode approach") based on clustering. In this work, we apply both approaches in a single- and multi-scale setting and compare the results in a multi-label histopathology classification task to show the promises and pitfalls of multi-scale analysis. Our work shows a consistent improvement in performance of the multi-scale models over single-scale ones. Using multiple instance learning and the barcode approach we achieved a 0.06 and 0.06 improvement in F1 score, respectively, highlighting the importance of combining multiple scales to integrate contextual and detailed information.

Electrical Compartmentalization in Neurons.

The dendritic tree of neurons plays an important role in information processing in the brain. While it is thought that dendrites require independent subunits to perform most of their computations, it is still not understood how they compartmentalize into functional subunits. Here, we show how these subunits can be deduced from the properties of dendrites. We devised a formalism that links the dendritic arborization to an impedance-based tree graph and show how the topology of this graph reveals independent subunits. This analysis reveals that cooperativity between synapses decreases slowly with increasing electrical separation and thus that few independent subunits coexist. We nevertheless find that balanced inputs or shunting inhibition can modify this topology and increase the number and size of the subunits in a context-dependent manner. We also find that this dynamic recompartmentalization can enable branch-specific learning of stimulus features. Analysis of dendritic patch-clamp recording experiments confirmed our theoretical predictions.

Design and implementation of multi-signal and time-varying neural reconstructions.

Several efficient procedures exist to digitally trace neuronal structure from light microscopy, and mature community resources have emerged to store, share, and analyze these datasets. In contrast, the quantification of intracellular distributions and morphological dynamics is not yet standardized. Current widespread descriptions of neuron morphology are static and inadequate for subcellular characterizations. We introduce a new file format to represent multichannel information as well as an open-source Vaa3D plugin to acquire this type of data. Next we define a novel data structure to capture morphological dynamics, and demonstrate its application to different time-lapse experiments. Importantly, we designed both innovations as judicious extensions of the classic SWC format, thus ensuring full back-compatibility with popular visualization and modeling tools. We then deploy the combined multichannel/time-varying reconstruction system on developing neurons in live Drosophila larvae by digitally tracing fluorescently labeled cytoskeletal components along with overall dendritic morphology as they changed over time. This same design is also suitable for quantifying dendritic calcium dynamics and tracking arbor-wide movement of any subcellular substrate of interest.

Cerebellar Nuclear Neurons Use Time and Rate Coding to Transmit Purkinje Neuron Pauses.

Neurons of the cerebellar nuclei convey the final output of the cerebellum to their targets in various parts of the brain. Within the cerebellum their direct upstream connections originate from inhibitory Purkinje neurons. Purkinje neurons have a complex firing pattern of regular spikes interrupted by intermittent pauses of variable length. How can the cerebellar nucleus process this complex input pattern? In this modeling study, we investigate different forms of Purkinje neuron simple spike pause synchrony and its influence on candidate coding strategies in the cerebellar nuclei. That is, we investigate how different alignments of synchronous pauses in synthetic Purkinje neuron spike trains affect either time-locking or rate-changes in the downstream nuclei. We find that Purkinje neuron synchrony is mainly represented by changes in the firing rate of cerebellar nuclei neurons. Pause beginning synchronization produced a unique effect on nuclei neuron firing, while the effect of pause ending and pause overlapping synchronization could not be distinguished from each other. Pause beginning synchronization produced better time-locking of nuclear neurons for short length pauses. We also characterize the effect of pause length and spike jitter on the nuclear neuron firing. Additionally, we find that the rate of rebound responses in nuclear neurons after a synchronous pause is controlled by the firing rate of Purkinje neurons preceding it.

A Sparse Reformulation of the Green's Function Formalism Allows Efficient Simulations of Morphological Neuron Models.

We prove that when a class of partial differential equations, generalized from the cable equation, is defined on tree graphs and the inputs are restricted to a spatially discrete, well chosen set of points, the Green's function (GF) formalism can be rewritten to scale as O(n) with the number n of inputs locations, contrary to the previously reported O(n(2)) scaling. We show that the linear scaling can be combined with an expansion of the remaining kernels as sums of exponentials to allow efficient simulations of equations from the aforementioned class. We furthermore validate this simulation paradigm on models of nerve cells and explore its relation with more traditional finite difference approaches. Situations in which a gain in computational performance is expected are discussed.

Centrosomin represses dendrite branching by orienting microtubule nucleation.

Neuronal dendrite branching is fundamental for building nervous systems. Branch formation is genetically encoded by transcriptional programs to create dendrite arbor morphological diversity for complex neuronal functions. In Drosophila sensory neurons, the transcription factor Abrupt represses branching via an unknown effector pathway. Targeted screening for branching-control effectors identified Centrosomin, the primary centrosome-associated protein for mitotic spindle maturation. Centrosomin repressed dendrite branch formation and was used by Abrupt to simplify arbor branching. Live imaging revealed that Centrosomin localized to the Golgi cis face and that it recruited microtubule nucleation to Golgi outposts for net retrograde microtubule polymerization away from nascent dendrite branches. Removal of Centrosomin enabled the engagement of wee Augmin activity to promote anterograde microtubule growth into the nascent branches, leading to increased branching. The findings reveal that polarized targeting of Centrosomin to Golgi outposts during elaboration of the dendrite arbor creates a local system for guiding microtubule polymerization.

Context-aware modeling of neuronal morphologies.

NEURONAL MORPHOLOGIES ARE PIVOTAL FOR BRAIN FUNCTIONING: physical overlap between dendrites and axons constrain the circuit topology, and the precise shape and composition of dendrites determine the integration of inputs to produce an output signal. At the same time, morphologies are highly diverse and variant. The variance, presumably, originates from neurons developing in a densely packed brain substrate where they interact (e.g., repulsion or attraction) with other actors in this substrate. However, when studying neurons their context is never part of the analysis and they are treated as if they existed in isolation. Here we argue that to fully understand neuronal morphology and its variance it is important to consider neurons in relation to each other and to other actors in the surrounding brain substrate, i.e., their context. We propose a context-aware computational framework, NeuroMaC, in which large numbers of neurons can be grown simultaneously according to growth rules expressed in terms of interactions between the developing neuron and the surrounding brain substrate. As a proof of principle, we demonstrate that by using NeuroMaC we can generate accurate virtual morphologies of distinct classes both in isolation and as part of neuronal forests. Accuracy is validated against population statistics of experimentally reconstructed morphologies. We show that context-aware generation of neurons can explain characteristics of variation. Indeed, plausible variation is an inherent property of the morphologies generated by context-aware rules. We speculate about the applicability of this framework to investigate morphologies and circuits, to classify healthy and pathological morphologies, and to generate large quantities of morphologies for large-scale modeling.

Spatially distributed dendritic resonance selectively filters synaptic input.

An important task performed by a neuron is the selection of relevant inputs from among thousands of synapses impinging on the dendritic tree. Synaptic plasticity enables this by strenghtening a subset of synapses that are, presumably, functionally relevant to the neuron. A different selection mechanism exploits the resonance of the dendritic membranes to preferentially filter synaptic inputs based on their temporal rates. A widely held view is that a neuron has one resonant frequency and thus can pass through one rate. Here we demonstrate through mathematical analyses and numerical simulations that dendritic resonance is inevitably a spatially distributed property; and therefore the resonance frequency varies along the dendrites, and thus endows neurons with a powerful spatiotemporal selection mechanism that is sensitive both to the dendritic location and the temporal structure of the incoming synaptic inputs.

Proposed evolutionary changes in the role of myelin.

Myelin is the multi-layered lipid sheet periodically wrapped around neuronal axons. It is most frequently found in vertebrates. Myelin allows for saltatory action potential (AP) conduction along axons. During this form of conduction, the AP travels passively along the myelin-covered part of the axon, and is recharged at the intermittent nodes of Ranvier. Thus, myelin can reduce the energy load needed and/or increase the speed of AP conduction. Myelin first evolved during the Ordovician period. We hypothesize that myelin's first role was mainly energy conservation. During the later "Mesozoic marine revolution," marine ecosystems changed toward an increase in marine predation pressure. We hypothesize that the main purpose of myelin changed from energy conservation to conduction speed increase during this Mesozoic marine revolution. To test this hypothesis, we optimized models of myelinated axons for a combination of AP conduction velocity and energy efficiency. We demonstrate that there is a trade-off between these objectives. We then compared the simulation results to empirical data and conclude that while the data are consistent with the theory, additional measurements are necessary for a complete evaluation of the proposed hypothesis.

The Green's function formalism as a bridge between single- and multi-compartmental modeling.

Neurons are spatially extended structures that receive and process inputs on their dendrites. It is generally accepted that neuronal computations arise from the active integration of synaptic inputs along a dendrite between the input location and the location of spike generation in the axon initial segment. However, many application such as simulations of brain networks use point-neurons-neurons without a morphological component-as computational units to keep the conceptual complexity and computational costs low. Inevitably, these applications thus omit a fundamental property of neuronal computation. In this work, we present an approach to model an artificial synapse that mimics dendritic processing without the need to explicitly simulate dendritic dynamics. The model synapse employs an analytic solution for the cable equation to compute the neuron's membrane potential following dendritic inputs. Green's function formalism is used to derive the closed version of the cable equation. We show that by using this synapse model, point-neurons can achieve results that were previously limited to the realms of multi-compartmental models. Moreover, a computational advantage is achieved when only a small number of simulated synapses impinge on a morphologically elaborate neuron. Opportunities and limitations are discussed.

Oscillatory activity, phase differences, and phase resetting in the inferior olivary nucleus.

The generation of temporal patterns is one of the most fascinating functions of the brain. Unlike the response to external stimuli temporal patterns are generated within the system and recalled for a specific use. To generate temporal patterns one needs a timing machine, a "master clock" that determines the temporal framework within which temporal patterns can be generated and implemented. Here we present the concept that in this putative "master clock" phase and frequency interact to generate temporal patterns. We define the requirements for a neuronal "master clock" to be both reliable and versatile. We introduce this concept within the inferior olive nucleus which at least by some scientists is regarded as the source of timing for cerebellar function. We review the basic properties of the subthreshold oscillation recorded from olivary neurons, analyze the phase relationships between neurons and demonstrate that the phase and onset of oscillation is tightly controlled by synaptic input. These properties endowed the olivary nucleus with the ability to act as a "master clock."

Self-referential forces are sufficient to explain different dendritic morphologies.

Dendritic morphology constrains brain activity, as it determines first which neuronal circuits are possible and second which dendritic computations can be performed over a neuron's inputs. It is known that a range of chemical cues can influence the final shape of dendrites during development. Here, we investigate the extent to which self-referential influences, cues generated by the neuron itself, might influence morphology. To this end, we developed a phenomenological model and algorithm to generate virtual morphologies, which are then compared to experimentally reconstructed morphologies. In the model, branching probability follows a Galton-Watson process, while the geometry is determined by "homotypic forces" exerting influence on the direction of random growth in a constrained space. We model three such homotypic forces, namely an inertial force based on membrane stiffness, a soma-oriented tropism, and a force of self-avoidance, as directional biases in the growth algorithm. With computer simulations we explored how each bias shapes neuronal morphologies. We show that based on these principles, we can generate realistic morphologies of several distinct neuronal types. We discuss the extent to which homotypic forces might influence real dendritic morphologies, and speculate about the influence of other environmental cues on neuronal shape and circuitry.

The generation of phase differences and frequency changes in a network model of inferior olive subthreshold oscillations.

It is commonly accepted that the Inferior Olive (IO) provides a timing signal to the cerebellum. Stable subthreshold oscillations in the IO can facilitate accurate timing by phase-locking spikes to the peaks of the oscillation. Several theoretical models accounting for the synchronized subthreshold oscillations have been proposed, however, two experimental observations remain an enigma. The first is the observation of frequent alterations in the frequency of the oscillations. The second is the observation of constant phase differences between simultaneously recorded neurons. In order to account for these two observations we constructed a canonical network model based on anatomical and physiological data from the IO. The constructed network is characterized by clustering of neurons with similar conductance densities, and by electrical coupling between neurons. Neurons inside a cluster are densely connected with weak strengths, while neurons belonging to different clusters are sparsely connected with stronger connections. We found that this type of network can robustly display stable subthreshold oscillations. The overall frequency of the network changes with the strength of the inter-cluster connections, and phase differences occur between neurons of different clusters. Moreover, the phase differences provide a mechanistic explanation for the experimentally observed propagating waves of activity in the IO. We conclude that the architecture of the network of electrically coupled neurons in combination with modulation of the inter-cluster coupling strengths can account for the experimentally observed frequency changes and the phase differences.

Wide-field motion integration in fly VS cells: insights from an inverse approach.

Fly lobula plate tangential cells are known to perform wide-field motion integration. It is assumed that the shape of these neurons, and in particular the shape of the subclass of VS cells, is responsible for this type of computation. We employed an inverse approach to investigate the morphology-function relationship underlying wide-field motion integration in VS cells. In the inverse approach detailed, model neurons are optimized to perform a predefined computation: here, wide-field motion integration. We embedded the model neurons to be optimized in a biologically plausible model of fly motion detection to provide realistic inputs, and subsequently optimized model neuron with and without active conductances (g(Na), g(K), g(K(Na))) along their dendrites to perform this computation. We found that both passive and active optimized model neurons perform well as wide-field motion integrators. In addition, all optimized morphologies share the same blueprint as real VS cells. In addition, we also found a recurring blueprint for the distribution of g(K) and g(Na) in the active models. Moreover, we demonstrate how this morphology and distribution of conductances contribute to wide-field motion integration. As such, by using the inverse approach we can predict the still unknown distribution of g(K) and g(Na) and their role in motion integration in VS cells.

A comparison of methods to determine neuronal phase-response curves.

The phase-response curve (PRC) is an important tool to determine the excitability type of single neurons which reveals consequences for their synchronizing properties. We review five methods to compute the PRC from both model data and experimental data and compare the numerically obtained results from each method. The main difference between the methods lies in the reliability which is influenced by the fluctuations in the spiking data and the number of spikes available for analysis. We discuss the significance of our results and provide guidelines to choose the best method based on the available data.

An inverse approach for elucidating dendritic function.

We outline an inverse approach for investigating dendritic function-structure relationships by optimizing dendritic trees for a priori chosen computational functions. The inverse approach can be applied in two different ways. First, we can use it as a "hypothesis generator" in which we optimize dendrites for a function of general interest. The optimization yields an artificial dendrite that is subsequently compared to real neurons. This comparison potentially allows us to propose hypotheses about the function of real neurons. In this way, we investigated dendrites that optimally perform input-order detection. Second, we can use it as a "function confirmation" by optimizing dendrites for functions hypothesized to be performed by classes of neurons. If the optimized, artificial, dendrites resemble the dendrites of real neurons the artificial dendrites corroborate the hypothesized function of the real neuron. Moreover, properties of the artificial dendrites can lead to predictions about yet unmeasured properties. In this way, we investigated wide-field motion integration performed by the VS cells of the fly visual system. In outlining the inverse approach and two applications, we also elaborate on the nature of dendritic function. We furthermore discuss the role of optimality in assigning functions to dendrites and point out interesting future directions.

Systematic mapping between dendritic function and structure.

For many classes of neurons, the relationship between computational function and dendritic morphology remains unclear. To gain insights into this relationship, we utilize an inverse approach in which we optimize model neurons with realistic morphologies and ion channel distributions (of I(KA) and I(CaT)) to perform a computational function. In this study, the desired function is input-order detection: neurons have to respond differentially to the arrival of two inputs in a different temporal order. There is a single free parameter in this function, namely, the time lag between the arrivals of the two inputs. Systematically varying this parameter allowed us to map one axis of function space to structure space. Because the function of the optimized model neurons is known with certainty, their thorough analysis provides insights into the relationship between the neurons' functions, morphologies, ion channel distributions, and electrophysiological dynamics. Finally, we discuss issues of optimality in nervous systems.

Non-parametric algorithmic generation of neuronal morphologies.

Generation algorithms allow for the generation of Virtual Neurons (VNs) from a small set of morphological properties. The set describes the morphological properties of real neurons in terms of statistical descriptors such as the number of branches and segment lengths (among others). The majority of reconstruction algorithms use the observed properties to estimate the parameters of a priori fixed probability distributions in order to construct statistical descriptors that fit well with the observed data. In this article, we present a non-parametric generation algorithm based on kernel density estimators (KDEs). The new algorithm is called KDE-NEURON: and has three advantages over parametric reconstruction algorithms: (1) no a priori specifications about the distributions underlying the real data, (2) peculiarities in the biological data will be reflected in the VNs, and (3) ability to reconstruct different cell types. We experimentally generated motor neurons and granule cells, and statistically validated the obtained results. Moreover, we assessed the quality of the prototype data set and observed that our generated neurons are as good as the prototype data in terms of the used statistical descriptors. The opportunities and limitations of data-driven algorithmic reconstruction of neurons are discussed.