Search

Neurons in the primary visual cortex (V1) are often classified as simple or complex cells, but it is debated whether they are discrete hierarchical classes of neurons or if they represent a continuum of variation within a single class of cells. Herein, we show that simple and complex cells may arise commonly from the feedforward projections from the retina. From analysis of the cortical receptive fields in cats, we show evidence that simple and complex cells originate from the periodic variation of ON–OFF segregation in the feedforward projection of retinal mosaics, by which they organize into periodic clusters in V1. From data in cats, we observed that clusters of simple and complex receptive fields correlate topographically with orientation maps, which supports our model prediction. Our results suggest that simple and complex cells are not two distinct neural populations but arise from common retinal afferents, simultaneous with orientation tuning.

Convolutional neural networks (CNNs) lack robustness to test image corruptions that are not seen during training. In this paper, we propose a new layer for CNNs that increases their robustness to several types of corruptions of the input images. We call it a ‘push–pull’ layer and compute its response as the combination of two half-wave rectified convolutions, with kernels of different size and opposite polarity. Its implementation is based on a biologically motivated model of certain neurons in the visual system that exhibit response suppression, known as push–pull inhibition. We validate our method by replacing the first convolutional layer of the LeNet, ResNet and DenseNet architectures with our push–pull layer. We train the networks on original training images from the MNIST and CIFAR data sets and test them on images with several corruptions, of different types and severities, that are unseen by the training process. We experiment with various configurations of the ResNet and DenseNet models on a benchmark test set with typical image corruptions constructed on the CIFAR test images. We demonstrate that our push–pull layer contributes to a considerable improvement in robustness of classification of corrupted images, while maintaining state-of-the-art performance on the original image classification task. We released the code and trained models at the url http://github.com/nicstrisc/Push-Pull-CNN-layer.

This paper presents a bioinspired visual saliency model. The end-stopping mechanism in the primary visual cortex is introduced in to extract features that represent contour information of latent salient objects such as corners, line intersections and line endpoints, which are combined together with brightness, color and orientation features to form the final saliency map. This model is an analog for the processing mechanism of visual signals along from retina, lateral geniculate nucleus(LGN)to primary visual cortex V1:Firstly, according to the characteristics of the retina and LGN, an input image is decomposed into brightness and opposite color channels; Then, the simple cell is simulated with 2D Gabor filters, and the amplitude of Gabor response is utilized to represent the response of complex cell; Finally, the response of an end-stopped cell is obtained by multiplying the response of two complex cells with different orientation, and outputs of V1 and LGN constitute a bottom-up saliency map. Experimental results on public datasets show that our model can accurately predict human fixations, and the performance achieves the state of the art of bottom-up saliency model.

This paper offers a theory for the origin of direction selectivity (DS) in the macaque primary visual cortex, V1. DS is essential for the perception of motion and control of pursuit eye movements. In the macaque visual pathway, neurons with DS first appear in V1, in the Simple cell population of the Magnocellular input layer 4C [Formula: see text]. The lateral geniculate nucleus (LGN) cells that project to these cortical neurons, however, are not direction selective. We hypothesize that DS is initiated in feed-forward LGN input, in the summed responses of LGN cells afferent to a cortical cell, and it is achieved through the interplay of 1) different visual response dynamics of ON and OFF LGN cells and 2) the wiring of ON and OFF LGN neurons to cortex. We identify specific temporal differences in the ON/OFF pathways that, together with item 2, produce distinct response time courses in separated subregions; analysis and simulations confirm the efficacy of the mechanisms proposed. To constrain the theory, we present data on Simple cells in layer 4C [Formula: see text] in response to drifting gratings. About half of the cells were found to have high DS, and the DS was broadband in spatial and temporal frequency (SF and TF). The proposed theory includes a complete analysis of how stimulus features such as SF and TF interact with ON/OFF dynamics and LGN-to-cortex wiring to determine the preferred direction and magnitude of DS.

The visual system is highly sensitive to spatial context for encoding luminance patterns. Context sensitivity inspired the proposal of many neural mechanisms for explaining the perception of luminance (brightness). Here we propose a novel computational model for estimating the brightness of many visual illusions. We hypothesize that many aspects of brightness can be explained by a dynamic filtering process that reduces the redundancy in edge representations on the one hand, while non-redundant activity is enhanced on the other. The dynamic filter is learned for each input image and implements context sensitivity. Dynamic filtering is applied to the responses of (model) complex cells in order to build a gain control map. The gain control map then acts on simple cell responses before they are used to create a brightness map via activity propagation. Our approach is successful in predicting many challenging visual illusions, including contrast effects, assimilation, and reverse contrast with the same set of model parameters., Author summary We hardly notice that what we see is often different from the physical world “outside” of the brain. This means that the visual experience that the brain actively constructs may be different from the actual physical properties of objects in the world. In this work, we propose a hypothesis about how the visual system of the brain may construct a representation for achromatic images. Since this process is not unambiguous, sometimes we notice “errors” in our perception, which cause visual illusions. The challenge for theorists, therefore, is to propose computational principles that recreate a large number of visual illusions and to explain why they occur. Notably, our proposed mechanism explains a broader set of visual illusions than any previously published proposal. We achieved this by trying to suppress predictable information. For example, if an image contains repetitive structures, then these structures are predictable and will be suppressed. In this way, non-predictable structures stand out. Corresponding mechanisms act as early as in the retina (which enhances luminance changes but suppresses uniform regions of luminance), and our computational model suggests that such mechanisms also might be used at subsequent stages in the visual system, where representations of perceived luminance (=brightness) are created.

There are two distinct classes of cells in the primary visual cortex (V1): simple cells and complex cells. One defining feature of complex cells is their spatial phase invariance; they respond strongly to oriented grating stimuli with a preferred orientation but with a wide range of spatial phases. A classical model of complete spatial phase invariance in complex cells is the energy model, in which the responses are the sum of the squared outputs of two linear spatially phase-shifted filters. However, recent experimental studies have shown that complex cells have a diverse range of spatial phase invariance and only a subset can be characterized by the energy model. While several models have been proposed to explain how complex cells could learn to be selective to orientation but invariant to spatial phase, most existing models overlook many biologically important details. We propose a biologically plausible model for complex cells that learns to pool inputs from simple cells based on the presentation of natural scene stimuli. The model is a three-layer network with rate-based neurons that describes the activities of LGN cells (layer 1), V1 simple cells (layer 2), and V1 complex cells (layer 3). The first two layers implement a recently proposed simple cell model that is biologically plausible and accounts for many experimental phenomena. The neural dynamics of the complex cells is modeled as the integration of simple cells inputs along with response normalization. Connections between LGN and simple cells are learned using Hebbian and anti-Hebbian plasticity. Connections between simple and complex cells are learned using a modified version of the Bienenstock, Cooper, and Munro (BCM) rule. Our results demonstrate that the learning rule can describe a diversity of complex cells, similar to those observed experimentally., Author summary Many cortical functions originate from the learning ability of the brain. How the properties of cortical cells are learned is vital for understanding how the brain works. There are many models that explain how V1 simple cells can be learned. However, how V1 complex cells are learned still remains unclear. In this paper, we propose a model of learning in complex cells based on the Bienenstock, Cooper, and Munro (BCM) rule. We demonstrate that properties of receptive fields of complex cells can be learned using this biologically plausible learning rule. Quantitative comparisons between the model and experimental data are performed. Results show that model complex cells can account for the diversity of complex cells found in experimental studies. In summary, this study provides a plausible explanation for how complex cells can be learned using biologically plausible plasticity mechanisms. Our findings help us to better understand biological vision processing and provide us with insights into the general signal processing principles that the visual cortex employs to process visual information.

This article gives an overview of a normative theory of visual receptive fields. We describe how idealized functional models of early spatial, spatio-chromatic and spatio-temporal receptive fields can be derived in a principled way, based on a set of axioms that reflect structural properties of the environment in combination with assumptions about the internal structure of a vision system to guarantee consistent handling of image representations over multiple spatial and temporal scales. Interestingly, this theory leads to predictions about visual receptive field shapes with qualitatively very good similarities to biological receptive fields measured in the retina, the LGN and the primary visual cortex (V1) of mammals., Receptive field; Functional model; Gaussian derivative; Scale covariance; Affine covariance; Galilean covariance; Temporal causality; Illumination invariance; Retina; LGN; Primary visual cortex; Simple cell; Double-opponent cell; Vision

This paper shows computationally and conceptually how gradients of signalling activity can be generated and dynamically maintained across a population of cells using very simple cell to cell communication rules. The rules work on the basis of cells regulating their production rate of a signalling molecule according to the production rates of their immediate neighbours. Highly stable, scale invariant signalling gradients can be formed across the population, with highest rates at the centre and lowest at the periphery.The cell to cell communication behaviour that causes gradient formation is first explained in a descriptive, thought experiment type manner. It is then defined more formally using a conceptual, mathematically discrete computational model, which provides a network or graph type framework in which it is easy to analyse and control discrete signals that are sent between neighbouring cells. This provides an intuitive method of explaining how the signalling gradient emerges as a result of local cell to cell communication. Finally, examples of gradient formation are shown using software implementations of the model.

A generalized compatible cell mapping (CCM) method is proposed in this paper to take advantages of the simple cell mapping (SCM) method, the generalized cell mapping (GCM) method together with a subdivision procedure. A coarse cell partition is first used to obtain a covering set of the global attractor. Then, a finer global attractor is obtained by the subdivision process. The probabilistic response of stochastic dynamic systems is obtained by the sparse matrix analysis algorithm applied to the covering set of the global attractor. Because the computational domain is the covering set of the global attractor rather than the whole state space, the numerical efficiency of the proposed method can be greatly improved as compared to the GCM. A three-dimensional and a four-dimensional dynamical system under Poisson white noise excitation are studied to demonstrate the effectiveness of the proposed method for the probabilistic response analysis. Monte Carlo simulations show a good agreement with the proposed method.

The visual system is highly sensitive to spatial context for encoding luminance patterns. Context sensitivity inspired the proposal of many neural mechanisms for explaining the perception of luminance (brightness). Here we propose a novel computational model for estimating the brightness of many visual illusions. We hypothesize that many aspects of brightness can be explained by a predictive coding mechanism, which reduces the redundancy in edge representations on the one hand, while non-redundant activity is enhanced on the other (response equalization). Response equalization is implemented with a dynamic filtering process, which (dynamically) adapts to each input image. Dynamic filtering is applied to the responses of complex cells in order to build a gain control map. The gain control map then acts on simple cell responses before they are used to create a brightness map via activity propagation. Our approach is successful in predicting many challenging visual illusions, including contrast effects, assimilation, and reverse contrast.Author summaryWe hardly notice that what we see is often different from the physical world “outside” of the brain. This means that the visual experience that the brain actively constructs may be different from the actual physical properties of objects in the world. In this work, we propose a hypothesis about how the visual system of the brain may construct a representation for achromatic images. Since this process is not unambiguous, sometimes we notice “errors” in our perception, which cause visual illusions. The challenge for theorists, therefore, is to propose computational principles that recreate a large number of visual illusions and to explain why they occur. Notably, our proposed mechanism explains a broader set of visual illusions than any previously published proposal. We achieved this by trying to suppress predictable information. For example, if an image contained repetitive structures, then these structures are predictable and would be suppressed. In this way, non-predictable structures stand out. Predictive coding mechanisms act as early as in the retina (which enhances luminance changes but suppresses uniform regions of luminance), and our computational model holds that this principle also acts at the next stage in the visual system, where representations of perceived luminance (brightness) are created.

Inhibition in thalamorecipient layer 4 simple cells of primary visual cortex is believed to play important roles in establishing visual response properties and integrating visual inputs across their receptive fields (RFs). Simple cell RFs are characterized by nonoverlapping, spatially restricted subregions in which visual stimuli can either increase or decrease the firing rate of the cell, depending on contrast. Inhibition is believed to be triggered exclusively from visual stimulation of individual RF subregions. However, this view is at odds with the known anatomy of layer 4 interneurons in visual cortex and differs from recent findings in mouse visual cortex. Here we show within vivointracellular recordings in cats that while excitation is restricted to RF subregions, inhibition spans the width of simple cell RFs. Consequently, excitatory stimuli within a subregion concomitantly drive excitation and inhibition. Furthermore, we found that the distribution of inhibition across the RF is stronger toward OFF subregions. This inhibitory OFF-subregion bias has a functional consequence on spatial integration of inputs across the RF. A model based on the known anatomy of layer 4 demonstrates that the known proportion and connectivity of inhibitory neurons in layer 4 of primary visual cortex is sufficient to explain broad inhibition with an OFF-subregion bias while generating a variety of phase relations, including antiphase, between excitation and inhibition in response to drifting gratings.SIGNIFICANCE STATEMENTThe wiring of excitatory and inhibitory neurons in cortical circuits is key to determining the response properties in sensory cortex. In the visual cortex, the first cells that receive visual input are simple cells in layer 4. The underlying circuitry responsible for the response properties of simple cells is not yet known. In this study, we challenge a long-held view concerning the pattern of inhibitory input and provide results that agree with current known anatomy. We show here that inhibition is evoked broadly across the receptive fields of simple cells, and we identify a surprising bias in inhibition within the receptive field. Our findings represent a step toward a unified view of inhibition across different species and sensory systems.

The two-dimensional Gabor function is adapted to natural image statistics, leading to a tractable probabilistic generative model that can be used to model simple cell receptive field profiles, or generate basis functions for sparse coding applications. Learning is found to be most pronounced in three Gabor function parameters representing the size and spatial frequency of the two-dimensional Gabor function and characterized by a nonuniform probability distribution with heavy tails. All three parameters are found to be strongly correlated, resulting in a basis of multiscale Gabor functions with similar aspect ratios and size-dependent spatial frequencies. A key finding is that the distribution of receptive-field sizes is scale invariant over a wide range of values, so there is no characteristic receptive field size selected by natural image statistics. The Gabor function aspect ratio is found to be approximately conserved by the learning rules and is therefore not well determined by natural image statistics. This allows for three distinct solutions: a basis of Gabor functions with sharp orientation resolution at the expense of spatial-frequency resolution, a basis of Gabor functions with sharp spatial-frequency resolution at the expense of orientation resolution, or a basis with unit aspect ratio. Arbitrary mixtures of all three cases are also possible. Two parameters controlling the shape of the marginal distributions in a probabilistic generative model fully account for all three solutions. The best-performing probabilistic generative model for sparse coding applications is found to be a gaussian copula with Pareto marginal probability density functions.

Transporting living complex cellular constructs through the mail while retaining their full viability and functionality is challenging. During this process, cells often suffer from exposure to suboptimal life-sustaining conditions (e.g. temperature, pH), as well as damage due to shear stress. We have developed a transport device for shipping intact cell/tissue constructs from one facility to another that overcomes these obstacles. Our transport device maintained three different cell lines (Caco2, A549, and HepG2 C3A) individually on transwell membranes with high viability (above 97%) for 48 h under simulated shipping conditions without an incubator. The device was also tested by actual overnight shipping of blood brain barrier constructs consisting of human induced pluripotent brain microvascular endothelial cells and rat astrocytes on transwell membranes to a remote facility (approximately 1200 miles away). The blood brain barrier constructs arrived with high cell viability and were able to regain full barrier integrity after equilibrating in the incubator for 24 h; this was assessed by the presence of continuous tight junction networks and in vivo-like values for trans-endothelial electrical resistance (TEER). These results demonstrated that our cell transport device could be a useful tool for long-distance transport of membrane-bound cell cultures and functional tissue constructs. Studies that involve various cell and tissue constructs, such as the "Multi-Organ-on-Chip" devices (where multiple microscale tissue constructs are integrated on a single microfluidic device) and studies that involve microenvironments where multiple tissue interactions are of interest, would benefit from the ability to transport or receive these constructs. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 33:1257-1266, 2017.

International audience; Microfluidic electrical impedance flow cytometry is now a well-known and established method for single-cell analysis. Given the richness of the information provided by impedance measurements, this non-invasive and label-free approach can be used in a wide field of applications ranging from simple cell counting to disease diagnostics. One of its major limitations is the variation of the impedance signal with the position of the cell in the sensing area. Indeed, identical particles traveling along different trajectories do not result in the same data. The positional dependence can be considered as a challenge for the accuracy of microfluidic impedance cytometers. On the other hand, it has recently been regarded by several groups as an opportunity to estimate the position of particles in the microchannel and thus take a further step in the logic of integrating sensors in so-called “Lab-on-a-chip” devices. This review provides a comprehensive overview of the physical grounds of the positional dependence of impedance measurements. Then, both the developed strategies to reduce position influence in impedance-based assays and the recent reported technologies exploiting that dependence for the integration of position detection in microfluidic devices are reviewed.

Segmentation of skin lesions is an important step in computer-aided diagnosis of melanoma; it is also a very challenging task due to fuzzy lesion boundaries and heterogeneous lesion textures. We present a fully automatic method for skin lesion segmentation based on deep fully convolutional networks (FCNs). We investigate a shallow encoding network to model clinically valuable prior knowledge, in which spatial filters simulating simple cell receptive fields function in the primary visual cortex (V1) is considered. An effective fusing strategy using skip connections and convolution operators is then leveraged to couple prior knowledge encoded via shallow network with hierarchical data-driven features learned from the FCNs for detailed segmentation of the skin lesions. To our best knowledge, this is the first time the domain-specific hand craft features have been built into a deep network trained in an end-to-end manner for skin lesion segmentation. The method has been evaluated on both ISBI 2016 and ISBI 2017 skin lesion challenge datasets. We provide comparative evidence to demonstrate that our newly designed network can gain accuracy for lesion segmentation by coupling the prior knowledge encoded by the shallow network with the deep FCNs. Our method is robust without the need for data augmentation or comprehensive parameter tuning, and the experimental results show great promise of the method with effective model generalization compared to other state-of-the-art-methods.

Intracellular drug delivery by rapid squeezing is one of the most recent and simple cell membrane disruption-mediated drug encapsulation approaches. In this method, cell membranes are perforated in a microfluidic setup due to rapid cell deformation during squeezing through constricted channels. While squeezing-based drug loading has been successful in loading drug molecules into various cell types, such as immune cells, cancer cells, and other primary cells, there is so far no comprehensive understanding of the pore opening mechanism on the cell membrane and the systematic analysis on how different channel geometries and squeezing speed influence drug loading. This article aims to develop a three-dimensional computational model to study the intracellular delivery for compound cells squeezing through microfluidic channels. The Lattice Boltzmann method, as the flow solver, integrated with a spring-connected network via frictional coupling, is employed to capture compound capsule dynamics over fast squeezing. The pore size is proportional to the local areal strain of triangular patches on the compound cell through mathematical correlations derived from molecular dynamics and coarse-grained molecular dynamics simulations. We quantify the drug concentration inside the cell cytoplasm by introducing a new mathematical model for passive diffusion after squeezing. Compared to the existing models, the proposed model does not have any empirical parameters that depend on operating conditions and device geometry. Since the compound cell model is new, it is validated by simulating a nucleated cell under a simple shear flow at different capillary numbers and comparing the results with other numerical models reported in literature. The cell deformation during squeezing is also compared with the pattern found from our compound cell squeezing experiment. Afterward, compound cell squeezing is modeled for different cell squeezing velocities, constriction lengths, and constriction widths. We reported the instantaneous cell center velocity, variations of axial and vertical cell dimensions, cell porosity, and normalized drug concentration to shed light on the underlying physics in fast squeezing-based drug delivery. Consistent with experimental findings in the literature, the numerical results confirm that constriction width reduction, constriction length enlargement, and average cell velocity promote intracellular drug delivery. The results show that the existence of the nucleus increases cell porosity and loaded drug concentration after squeezing. Given geometrical parameters and cell average velocity, the maximum porosity is achieved at three different locations: constriction entrance, constriction middle part, and outside the constriction. Our numerical results provide reasonable justifications for experimental findings on the influences of constriction geometry and cell velocity on the performance of cell-squeezing delivery. We expect this model can help design and optimize squeezing-based cargo delivery.

When a dynamical system has a complex structure of fixed points, periodic cycles or even chaotic attractors, Cell Mapping methods are excellent tools to discover and thoroughly analyse all features in the state space. These methods discretize a region of the state space into cells and examine the dynamics in the cell state space. By determining one or more image cells for each cell, the global behaviour within the region can be quickly determined. In the simplest case – Simple Cell Mapping (SCM) method – only one image corresponds to a cell and usually a rectangular grid of cells is used. In typical applications the grid of cells is refined at specific locations. This paper, however, introduces a different approach, which is useful to expand the analysed state space region to include all features which properly characterize the global dynamics of the system. Instead of refining the initial cell state space, we start with a small initial state space region, analyse other interesting regions of the state space and incorporate them into a cluster of cell mapping solutions. By this approach, trajectories escaping the original state space region can be followed automatically and additional objects in the state space can be discovered. To illustrate the benefits of the method, we present the exploration of the phase-space of the micro-chaos map – a simple model of digitally controlled systems.

Summary Dynamic actin networks are excitable. In migrating cells, feedback loops can amplify stochastic fluctuations in actin dynamics, often resulting in traveling waves of protrusion. The precise contributions of various molecular and mechanical interactions to wave generation have been difficult to disentangle, in part due to complex cellular morphodynamics. Here we used a relatively simple cell type—the fish epithelial keratocyte—to define a set of mechanochemical feedback loops underlying actin network excitability and wave generation. Although keratocytes are normally characterized by the persistent protrusion of a broad leading edge, increasing cell-substrate adhesion strength results in waving protrusion of a short leading edge. We show that protrusion waves are due to fluctuations in actin polymerization rates and that overexpression of VASP, an actin anti-capping protein that promotes actin polymerization, switches highly adherent keratocytes from waving to persistent protrusion. Moreover, VASP localizes both to adhesion complexes and to the leading edge. Based on these results, we developed a mathematical model for protrusion waves in which local depletion of VASP from the leading edge by adhesions—along with lateral propagation of protrusion due to the branched architecture of the actin network and negative mechanical feedback from the cell membrane—results in regular protrusion waves. Consistent with our model simulations, we show that VASP localization at the leading edge oscillates, with VASP leading-edge enrichment greatest just prior to protrusion initiation. We propose that the mechanochemical feedbacks underlying wave generation in keratocytes may constitute a general module for establishing excitable actin dynamics in other cellular contexts.

This paper presents a study of the multi-objective optimal design of a sliding mode control for an under-actuated nonlinear system with the parallel simple cell mapping method. The multi-objective optimal design of the sliding mode control involves six design parameters and five objective functions. The parallel simple cell mapping method finds the Pareto set and Pareto front efficiently. The parallel computing is done on a graphics processing unit. Numerical simulations and experiments are done on a rotary flexible arm system. The results show that the proposed multi-objective designs are quite effective.

The contour of an arbitrary figure can be represented as a group of circles of curvature in contact with it, with each curvature circle represented by its center OC and radius r. We propose a series of cell models for detecting this circle, which is composed of a lateral geniculate nucleus (LGN) cell, nondirectionally selective (NDS) simple cell, and curvature-circle detection cell (CDC). The LGN and NDS simple cells were previously modeled. The CDC has been modeled as follows. Each tangent in contact with this circle is detected by an NDS simple cell that performs the Hough transformation of LGN cell responses, and then this tangent is transformed to a three-dimensional (3D) normal line in a CDC column. This transformation has been named a 3D normal-line transform. Performing this transformation for all tangents causes a CDC at the intersection of these normal lines to fire most intensively, and thus the OC and r of the circle is detected as the coordinates of this intersection. Therefore, the CDC has been modeled as this 3D normal-line transform. Based on this CDC, we model two types of constancy CDC: a position-invariant CDC and a curvature-invariant CDC. These three types of CDC reflect the response to various stimuli in actual area V4 cells. In order to validate these CDC types neurophysiologically, we propose an experimental method using microelectrodes. Cell models previously reported correspond to this hierarchy: the S1, S2, and C2 cells correspond to the NDS simple cell, CDC, and position-invariant CDC, respectively., Cell model, Curvature-circle detection, 3D normal-line transform, Column, Coarse-to-fine extraction, Cell-array conversion, Shape recognition, Information systems, Behavioral neuroscience, Nervous system, Cognition, Consciousness, Emotion, Systems neuroscience, Mathematical biosciences.

Increasing evidence supports the hypothesis that the visual system employs a sparse code to represent visual stimuli, where information is encoded in an efficient way by a small population of cells that respond to sensory input at a given time. This includes simple cells in primary visual cortex (V1), which are defined by their linear spatial integration of visual stimuli. Various models of sparse coding have been proposed to explain physiological phenomena observed in simple cells. However, these models have usually made the simplifying assumption that inputs to simple cells already incorporate linear spatial summation. This overlooks the fact that these inputs are known to have strong non-linearities such the separation of ON and OFF pathways, or separation of excitatory and inhibitory neurons. Consequently these models ignore a range of important experimental phenomena that are related to the emergence of linear spatial summation from non-linear inputs, such as segregation of ON and OFF sub-regions of simple cell receptive fields, the push-pull effect of excitation and inhibition, and phase-reversed cortico-thalamic feedback. Here, we demonstrate that a two-layer model of the visual pathway from the lateral geniculate nucleus to V1 that incorporates these biological constraints on the neural circuits and is based on sparse coding can account for the emergence of these experimental phenomena, diverse shapes of receptive fields and contrast invariance of orientation tuning of simple cells when the model is trained on natural images. The model suggests that sparse coding can be implemented by the V1 simple cells using neural circuits with a simple biologically plausible architecture.

Background Accurate prediction of anticancer drug responses in cell lines is a crucial step to accomplish the precision medicine in oncology. Although many popular computational models have been proposed towards this non-trivial issue, there is still room for improving the prediction performance by combining multiple types of genome-wide molecular data. Results We first demonstrated an observation on the CCLE and GDSC datasets, i.e., genetically similar cell lines always exhibit higher response correlations to structurally related drugs. Based on this observation we built a cell line-drug complex network model, named CDCN model. It captures different contributions of all available cell line-drug responses through cell line similarities and drug similarities. We executed anticancer drug response prediction on CCLE and GDSC independently. The result is significantly superior to that of some existing studies. More importantly, our model could predict the response of new drug to new cell line with considerable performance. We also divided all possible cell lines into “sensitive” and “resistant” groups by their response values to a given drug, the prediction accuracy, sensitivity, specificity and goodness of fit are also very promising. Conclusion CDCN model is a comprehensive tool to predict anticancer drug responses. Compared with existing methods, it is able to provide more satisfactory prediction results with less computational consumption. Electronic supplementary material The online version of this article (10.1186/s12859-019-2608-9) contains supplementary material, which is available to authorized users.

An indicator of a passive biped walker’s global stability is its domain of attraction, which is usually estimated by the simple cell mapping method. It needs to calculate a large number of cells’ Poincare mapping result in the estimating process. However, the Poincare mapping is usually computationally expensive and time-consuming due to the complex dynamical equation of the passive biped walker. How to estimate the domain of attraction efficiently and reliably is a problem to be solved. Based on the simple cell mapping method, an improved method is proposed to solve it. The proposed method uses the multiple iteration algorithm to calculate a stable domain of attraction and effectively decreases the total number of Poincare mappings. Through the simulation of the simplest passive biped walker, the improved method can obtain the same domain of attraction as that calculated using the simple cell mapping method and reduce calculation time significantly. Furthermore, this improved method not only proposes a way of rapid estimating the domain of attraction, but also provides a feasible tool for selecting the domain of interest and its discretization level.

Neurons that signal the orientation of edges within the visual field have been widely studied in primary visual cortex. Much less is known about the mechanisms of orientation selectivity that arise earlier in the visual stream. Here we examine the synaptic and morphological properties of a subtype of orientation-selective ganglion cell in the rabbit retina. The receptive field has an excitatory ON center, flanked by excitatory OFF regions, a structure similar to simple cell receptive fields in primary visual cortex. Examination of the light-evoked postsynaptic currents in these ON-type orientation-selective ganglion cells (ON-OSGCs) reveals that synaptic input is mediated almost exclusively through the ON pathway. Orientation selectivity is generated by larger excitation for preferred relative to orthogonal stimuli, and conversely larger inhibition for orthogonal relative to preferred stimuli. Excitatory orientation selectivity arises in part from the morphology of the dendritic arbors. Blocking GABAAreceptors reduces orientation selectivity of the inhibitory synaptic inputs and the spiking responses. Negative contrast stimuli in the flanking regions produce orientation-selective excitation in part by disinhibition of a tonic NMDA receptor-mediated input arising from ON bipolar cells. Comparison with earlier studies of OFF-type OSGCs indicates that diverse synaptic circuits have evolved in the retina to detect the orientation of edges in the visual input.SIGNIFICANCE STATEMENTA core goal for visual neuroscientists is to understand how neural circuits at each stage of the visual system extract and encode features from the visual scene. This study documents a novel type of orientation-selective ganglion cell in the retina and shows that the receptive field structure is remarkably similar to that of simple cells in primary visual cortex. However, the data indicate that, unlike in the cortex, orientation selectivity in the retina depends on the activity of inhibitory interneurons. The results further reveal the physiological basis for feature detection in the visual system, elucidate the synaptic mechanisms that generate orientation selectivity at an early stage of visual processing, and illustrate a novel role for NMDA receptors in retinal processing.

The response properties of neurons in the early stages of the visual system can be described using the rectified responses of a set of self-similar, spatially shifted linear filters. In macaque primary visual cortex (V1), simple cell responses can be captured with a single filter, whereas complex cells combine a set of filters, creating position invariance. These filters cannot be estimated using standard methods, such as spike-triggered averaging. Subspace methods like spike-triggered covariance can recover multiple filters but require substantial amounts of data, and recover an orthogonal basis for the subspace in which the filters reside, rather than the filters themselves. Here, we assume a linear-nonlinear-linear-nonlinear (LN-LN) cascade model in which the first LN stage consists of shifted (“convolutional”) copies of a single filter, followed by a common instantaneous nonlinearity. We refer to these initial LN elements as the “subunits” of the receptive field, and we allow two independent sets of subunits, each with its own filter and nonlinearity. The second linear stage computes a weighted sum of the subunit responses and passes the result through a final instantaneous nonlinearity. We develop a procedure to directly fit this model to electrophysiological data. When fit to data from macaque V1, the subunit model significantly outperforms three alternatives in terms of cross-validated accuracy and efficiency, and provides a robust, biologically plausible account of receptive field structure for all cell types encountered in V1.SIGNIFICANCE STATEMENTWe present a new subunit model for neurons in primary visual cortex that significantly outperforms three alternative models in terms of cross-validated accuracy and efficiency, and provides a robust and biologically plausible account of the receptive field structure in these neurons across the full spectrum of response properties.

In establishing the sources of data variability within sedimentation velocity analysis in the analytical ultracentrifuge and their relative importance, recent studies have demonstrated that alignment of the sample cells to the centre of rotation is the most significant contributing factor to overall variability, particularly for the characterisation of low levels of protein aggregation. Accurate mechanical and optical alignment tools have been recently designed. In this study, we (1) confirm the effect of misalignment observed by others on the estimated amounts of bovine serum albumin (BSA) monomer and dimer, and the sedimentation coefficient value for the BSA dimer; and (2) demonstrate the high performance of a mechanical alignment tool and the usefulness of a simple and complementary enhanced manual alignment protocol which should be useful for situations where these tools are not available.

Deep learning has thoroughly changed the field of image analysis yielding impressive results whenever enough annotated data can be gathered. While partial annotation can be very fast, manual segmentation of 3D biological structures is tedious and error-prone. Additionally, high-level shape concepts such as topology or boundary smoothness are hard if not impossible to encode in Feedforward Neural Networks. Here we present a modular strategy for the accurate segmentation of neural cell bodies from light-sheet microscopy combining mixed-scale convolutional neural networks and topology-preserving geometric deformable models. We show that the network can be trained efficiently from simple cell centroid annotations, and that the final segmentation provides accurate cell detection and smooth segmentations that do not introduce further cell splitting or merging.

Some neurons in early visual cortex are highly selective for the position of oriented edges in their receptive fields (simple cells), whereas others are largely position insensitive (complex cells). These characteristics are reflected in their sensitivity to the spatial phase of moving sine-wave gratings: simple cell responses oscillate at the fundamental frequency of the stimulus, whereas this is less so for complex cells. In primates, when assessed at high stimulus contrast, simple cells and complex cells are roughly equally represented in the first visual cortical area, V1, whereas in the second visual area, V2, the majority of cells are complex. Recent evidence has shown that phase sensitivity of complex cells is contrast dependent. This has led to speculation that reduced contrast may lead to changes in the spatial structure of receptive fields, perhaps due to changes in how feedforward and recurrent signals interact. Given the substantial interconnections between V1 and V2 and recent evidence for the emergence of unique functional capacity in V2, we assess the relationship between contrast and phase sensitivity in the two brain regions. We show that a substantial proportion of complex cells in macaque V1 exhibit significant increases in phase sensitivity at low contrast, whereas this is rarely observed in V2. Our results support a degree of hierarchical processing from V1 to V2 with the differences possibly relating to the fact that V1 combines both subcortical and cortical input, whereas V2 receives input purely from cortical circuits.

During cleavage, different cellular processes cause the zygote to become partitioned into a set of cells with a specific spatial arrangement. These processes include the orientation of cell division according to: an animal-vegetal gradient; the main axis (Hertwig’s rule) of the cell; and the contact areas between cells or the perpendicularity between consecutive cell divisions (Sachs’ rule). Cell adhesion and cortical rotation have also been proposed to be involved in spiral cleavage.We use a computational model of cell and tissue biomechanics to account for the different existing hypotheses about how the specific spatial arrangement of cells in spiral cleavage arises during development. Cell polarization by an animal-vegetal gradient, a bias to perpendicularity between consecutive cell divisions (Sachs’ rule), cortical rotation and cell adhesion, when combined, reproduce the spiral cleavage, whereas other combinations of processes cannot. Specifically, cortical rotation is necessary at the 8-cell stage to direct all micromeres in the same direction. By varying the relative strength of these processes, we reproduce the spatial arrangement of cells in the blastulae of seven different invertebrate species., Ministerio de Ciencia y Tecnologı́a (BFU2010-17044 to I.S.-C., BES2011-046641 to M.B.-U. and BES 2012-052214 to M.T.-G.), by the Generalitat de Catalunya (2013FI-B00439 to M.M.-R.), and by Universidad Autónoma de Madrid and Ministerio de Ciencia y Tecnologı́a (CGL2011-29916 to C.G.)

Background Recent developments in droplet-based microfluidics allow the transcriptional profiling of thousands of individual cells in a quantitative, highly parallel and cost-effective way. A critical, often limiting step is the preparation of cells in an unperturbed state, not altered by stress or ageing. Other challenges are rare cells that need to be collected over several days or samples prepared at different times or locations. Methods Here, we used chemical fixation to address these problems. Methanol fixation allowed us to stabilise and preserve dissociated cells for weeks without compromising single-cell RNA sequencing data. Results By using mixtures of fixed, cultured human and mouse cells, we first showed that individual transcriptomes could be confidently assigned to one of the two species. Single-cell gene expression from live and fixed samples correlated well with bulk mRNA-seq data. We then applied methanol fixation to transcriptionally profile primary cells from dissociated, complex tissues. Low RNA content cells from Drosophila embryos, as well as mouse hindbrain and cerebellum cells prepared by fluorescence-activated cell sorting, were successfully analysed after fixation, storage and single-cell droplet RNA-seq. We were able to identify diverse cell populations, including neuronal subtypes. As an additional resource, we provide 'dropbead', an R package for exploratory data analysis, visualization and filtering of Drop-seq data. Conclusions We expect that the availability of a simple cell fixation method will open up many new opportunities in diverse biological contexts to analyse transcriptional dynamics at single-cell resolution. Electronic supplementary material The online version of this article (doi:10.1186/s12915-017-0383-5) contains supplementary material, which is available to authorized users.

Through statistical analysis of datasets describing single cell shape following systematic gene depletion, we have found that the morphological landscapes explored by cells are composed of a small number of attractor states. We propose that the topology of these landscapes is in large part determined by cell-intrinsic factors, such as biophysical constraints on cytoskeletal organization, and reflects different stable signaling and/or transcriptional states. Cell-extrinsic factors act to determine how cells explore these landscapes, and the topology of the landscapes themselves. Informational stimuli primarily drive transitions between stable states by engaging signaling networks, while mechanical stimuli tune, or even radically alter, the topology of these landscapes. As environments fluctuate, the topology of morphological landscapes explored by cells dynamically adapts to these fluctuations. Finally we hypothesize how complex cellular and tissue morphologies can be generated from a limited number of simple cell shapes.

To model the responses of neurons in the early visual system, at least three basic components are required: a receptive field, a normalization term, and a specification of encoding noise. Here, we examine how the receptive field, the normalization factor, and the encoding noise affect the drive to model-neuron responses when stimulated with natural images. We show that when these components are modeled appropriately, the response drives elicited by natural stimuli are Gaussian-distributed and scale invariant, and very nearly maximize the sensitivity (d′) for natural-image discrimination. We discuss the statistical models of natural stimuli that can account for these response statistics, and we show how some commonly used modeling practices may distort these results. Finally, we show that normalization can equalize important properties of neural response across different stimulus types. Specifically, narrowband (stimulus- and feature-specific) normalization causes model neurons to yield Gaussian response-drive statistics when stimulated with natural stimuli, 1/f noise stimuli, and white-noise stimuli. The current work makes recommendations for best practices and lays a foundation, grounded in the response statistics to natural stimuli, upon which to build principled models of more complex visual tasks.

Background: Multiplexed immunofluorescence (mIF) has the potential to revolutionize immuno-oncology and pathology research as it enables the identification of complex cell phenotypes and their potential interactions in the tumor microenvironment (TME). But for a whole slide image with millions of cells, as we increase the number of biomarkers imaged for every cell, the complexity of the data analysis and visualization task increases exponentially. For n markers, a total of 2n phenotypes are possible (e.g. 10 markers have 1024 potential phenotypes). To address this problem and reveal the biologically relevant information embedded in the data, we have developed software tools to reduce the complexity, visualize, and quantify spatial distributions of cells across the full spectrum of possible phenotypes. Methods: Here we present results using two different methods. The first is an image processing technique called Phenotypic Surface Density Mapping (PSDM), that produces not only true surface density images of each phenotype (cells/µm2), but also surface density images that quantify a variety of other statistics such as the level of expression (intensity) of key markers, inter-phenotype nearest neighbor distance maps, and maps of cell size/morphology. Some important features of these surface density maps are that they are quantitatively robust, have real physical units (e.g. cells/µm2 or intensity/µm2), and they are generated in an unbiased fashion to reveal information about every possible phenotype. The second analysis method, dimensionality reduction, exploits a new technique called Uniform Manifold Approximation and Projection (UMAP), reducing dozens or even hundreds of dimensions for millions of cells to a simple 2D scatter plot. We have developed interactive software that displays the UMAP for a slide and allows the user to select a given cell or region of cells to view closeup images of each cell and statistics about the collection. Results: Examples of the surface density maps provide insights into mapping the complexity of the TME. We assess the results on deidentified samples by comparison with both human generated labels (pathology review) for individual cells and with automatically generated labels (software review). We show how these tools can be used to both identify tumors and quantify the level of activity in different tumor regions. We demonstrate how increasing the level of multiplexing allows one to differentiate subtle variability and separate subclasses of cells from each other. Conclusions: Multiplexed data brings valuable information about the TME but much of this information is inaccessible by simply viewing the captured images or performing simple cell counting alone. To address this problem, we demonstrate two new software tools, PSDM and UMAP that preserve and quantify the spatial information of the underlying biology and provide this analysis for all possible phenotypes. Citation Format: Douglas Wood, Bonnie Phillips, Courtney Hebert, Aditi Sharma, Jamie Buell, Sean Downing. Understanding the TME: Advanced analysis and visualization of multiplexed fluorescence images [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 1677.

In the classic model of the primary visual cortex, upper-layer complex cells are driven by feedforward inputs from layer 4 simple cells. Based on spike cross-correlation, previousin vivowork has suggested that this connection is strong and dense, with a high probability of connection (50%) and significant strength in connected pairs. A much sparser projection has been found in brain slices, however, with the probability of layer 4 cells connecting to layer 2/3 cells being relatively low (10%). Here, we explore this connectionin vivoin the cat primary visual cortex by recording simultaneously spikes of layer 4 simple cells and the membrane potential (Vm) of layer 2/3 complex cells. By triggering the average of the complex cell'sVmon the spikes of the simple cell (Vm-STA), we found functional coupling to be very common during visual stimulation: the simple cell's spikes tended to occur near the troughs of the complex cell'sVmfluctuations and were, on average, followed by a significant (∼1 mV) fast-rising (10 ms) depolarization in the complex cell. In the absence of visual stimulation, however, when single simple cells were activated electrically through the recording electrode, no significant depolarization, or at most a very weak input (0.1–0.2 mV), was detected in the complex cell. We suggest that the functional coupling observed during visual stimulation arises from coordinated or nearly synchronous activity among a large population of simple cells, only a small fraction of which are presynaptic to the recorded complex cell.

A receptive field constitutes a region in the visual field where a visual cell or a visual operator responds to visual stimuli. This paper presents a theory for what types of receptive field profiles can be regarded as natural for an idealized vision system, given a set of structural requirements on the first stages of visual processing that reflect symmetry properties of the surrounding world. These symmetry properties include (i) covariance properties under scale changes, affine image deformations, and Galilean transformations of space–time as occur for real-world image data as well as specific requirements of (ii) temporal causality implying that the future cannot be accessed and (iii) a time-recursive updating mechanism of a limited temporal buffer of the past as is necessary for a genuine real-time system. Fundamental structural requirements are also imposed to ensure (iv) mutual consistency and a proper handling of internal representations at different spatial and temporal scales. It is shown how a set of families of idealized receptive field profiles can be derived by necessity regarding spatial, spatio-chromatic, and spatio-temporal receptive fields in terms of Gaussian kernels, Gaussian derivatives, or closely related operators. Such image filters have been successfully used as a basis for expressing a large number of visual operations in computer vision, regarding feature detection, feature classification, motion estimation, object recognition, spatio-temporal recognition, and shape estimation. Hence, the associated so-called scale-space theory constitutes a both theoretically well-founded and general framework for expressing visual operations. There are very close similarities between receptive field profiles predicted from this scale-space theory and receptive field profiles found by cell recordings in biological vision. Among the family of receptive field profiles derived by necessity from the assumptions, idealized models with very good qualitative agreement are obtained for (i) spatial on-center/off-surround and off-center/on-surround receptive fields in the fovea and the LGN, (ii) simple cells with spatial directional preference in V1, (iii) spatio-chromatic double-opponent neurons in V1, (iv) space–time separable spatio-temporal receptive fields in the LGN and V1, and (v) non-separable space–time tilted receptive fields in V1, all within the same unified theory. In addition, the paper presents a more general framework for relating and interpreting these receptive fields conceptually and possibly predicting new receptive field profiles as well as for pre-wiring covariance under scaling, affine, and Galilean transformations into the representations of visual stimuli. This paper describes the basic structure of the necessity results concerning receptive field profiles regarding the mathematical foundation of the theory and outlines how the proposed theory could be used in further studies and modelling of biological vision. It is also shown how receptive field responses can be interpreted physically, as the superposition of relative variations of surface structure and illumination variations, given a logarithmic brightness scale, and how receptive field measurements will be invariant under multiplicative illumination variations and exposure control mechanisms. QC 20131210