Automatic dendritic spine analysis in two-photon laser scanning microscopy images

Utilizing 2D-region-based CNNs for automatic dendritic spine detection in 3D live cell imaging

Dendritic spines are considered a morphological proxy for excitatory synapses, rendering them a target of many different lines of research. Over recent years, it has become possible to simultaneously image large numbers of dendritic spines in 3D volumes of neural tissue. In contrast, currently no automated method for 3D spine detection exists that comes close to the detection performance reached by human experts. However, exploiting such datasets requires new tools for the fully automated detection and analysis of large numbers of spines. Here, we developed an efficient analysis pipeline to detect large numbers of dendritic spines in volumetric fluorescence imaging data acquired by two-photon imaging in vivo. The core of our pipeline is a deep convolutional neural network that was pretrained on a general-purpose image library and then optimized on the spine detection task. This transfer learning approach is data efficient while achieving a high detection precision. To train and validate the model we generated a labeled dataset using five human expert annotators to account for the variability in human spine detection. The pipeline enables fully automated dendritic spine detection reaching a performance slightly below that of the human experts. Our method for spine detection is fast, accurate and robust, and thus well suited for large-scale datasets with thousands of spines. The code is easily applicable to new datasets, achieving high detection performance, even without any retraining or adjustment of model parameters.

Current Best Practices for Analysis of Dendritic Spine Morphology and Number in Neurodevelopmental Disorder Research

Quantitative methods for assessing neural anatomy have rapidly evolved in neuroscience and provide important insights into brain health and function. However, as new techniques develop, it is not always clear when and how each may be used to answer specific scientific questions posed. Dendritic spines, which are often indicative of synapse formation and neural plasticity, have been implicated across many brain regions in neurodevelopmental disorders as a marker for neural changes reflecting neural dysfunction or alterations. In this Perspective we highlight several techniques for staining, imaging, and quantifying dendritic spines as well as provide a framework for avoiding potential issues related to pseudoreplication. This framework illustrates how others may apply the most rigorous approaches. We consider the cost-benefit analysis of the varied techniques, recognizing that the most sophisticated equipment may not always be necessary for answering some research questions. Together, we hope this piece will help researchers determine the best strategy toward using the ever-growing number of techniques available to determine neural changes underlying dendritic spine morphology in health and neurodevelopmental disorders.

Fear memory-associated synaptic and mitochondrial changes revealed by deep learning-based processing of electron microscopy data

Serial section electron microscopy (ssEM) can provide comprehensive 3D ultrastructural information of the brain with exceptional computational cost. Targeted reconstruction of subcellular structures from ssEM datasets is less computationally demanding but still highly informative. We thus developed a region-CNN-based deep learning method to identify, segment, and reconstruct synapses and mitochondria to explore the structural plasticity of synapses and mitochondria in the auditory cortex of mice subjected to fear conditioning. Upon reconstructing over 135,000 mitochondria and 160,000 synapses, we find that fear conditioning significantly increases the number of mitochondria but decreases their size and promotes formation of multi-contact synapses, comprising a single axonal bouton and multiple postsynaptic sites from different dendrites. Modeling indicates that such multi-contact configuration increases the information storage capacity of new synapses by over 50%. With high accuracy and speed in reconstruction, our method yields structural and functional insight into cellular plasticity associated with fear learning.

Recent advances in computational methods for measurement of dendritic spines imaged by light microscopy

Dendritic spines are small protrusions that receive most of the excitatory inputs to the pyramidal neurons in the neocortex and the hippocampus. Excitatory neural circuits in the neocortex and hippocampus are important for experience-dependent changes in brain functions, including postnatal sensory refinement and memory formation. Several lines of evidence indicate that synaptic efficacy is correlated with spine size and structure. Hence, precise and accurate measurement of spine morphology is important for evaluation of neural circuit function and plasticity. Recent advances in light microscopy and image analysis techniques have opened the way toward a full description of spine nanostructure. In addition, large datasets of spine nanostructure can be effectively analyzed using machine learning techniques and other mathematical approaches, and recent advances in super-resolution imaging allow researchers to analyze spine structure at an unprecedented level of precision. This review summarizes computational methods that can effectively identify, segment and quantitate dendritic spines in either 2D or 3D imaging. Nanoscale analysis of spine structure and dynamics, combined with new mathematical approaches, will facilitate our understanding of spine functions in physiological and pathological conditions.

Automated dendritic spine detection using convolutional neural networks on maximum intensity projected microscopic volumes

BACKGROUND

Dendritic spines are structural correlates of excitatory synapses in the brain. Their density and structure are shaped by experience, pointing to their role in memory encoding. Dendritic spine imaging, followed by manual analysis, is a primary way to study spines. However, an approach that analyses dendritic spines images in an automated and unbiased manner is needed to fully capture how spines change with normal experience, as well as in disease.

NEW METHOD

We propose an approach based on fully convolutional neural networks (FCNs) to detect dendritic spines in two-dimensional maximum-intensity projected images from confocal fluorescent micrographs. We experiment on both fractionally strided convolution and efficient sub-pixel convolutions. Dendritic spines far from the dendritic shaft are pruned by extraction of the shaft to reduce false positives. Performance of the proposed method is evaluated by comparing predicted spine positions to those manually marked by experts.

RESULTS

The averaged distance between predicted and manually annotated spines is 2.81 ± 2.63 pixels (0.082 ± 0.076 microns) and 2.87 ± 2.33 pixels (0.084 ± 0.068 microns) based on two different experts. FCN-based detection achieves F scores > 0.80 for both sets of expert annotations.

COMPARISON WITH EXISTING METHODS

Our method significantly outperforms two well-known software, NeuronStudio and Neurolucida (p-value < 0.02).

CONCLUSIONS

FCN architectures used in this work allow for automated dendritic spine detection. Superior outcomes are possible even with small training data-sets. The proposed method may generalize to other datasets on larger scales.

Image Informatics Strategies for Deciphering Neuronal Network Connectivity

Brain function relies on an intricate network of highly dynamic neuronal connections that rewires dramatically under the impulse of various external cues and pathological conditions. Amongst the neuronal structures that show morphological plasticity are neurites, synapses, dendritic spines and even nuclei. This structural remodelling is directly connected with functional changes such as intercellular communication and the associated calcium bursting behaviour. In vitro cultured neuronal networks are valuable models for studying these morpho-functional changes. Owing to the automation and standardization of both image acquisition and image analysis, it has become possible to extract statistically relevant readouts from such networks. Here, we focus on the current state-of-the-art in image informatics that enables quantitative microscopic interrogation of neuronal networks. We describe the major correlates of neuronal connectivity and present workflows for analysing them. Finally, we provide an outlook on the challenges that remain to be addressed, and discuss how imaging algorithms can be extended beyond in vitro imaging studies.

Oriented Markov random field based dendritic spine segmentation for fluorescence microscopy images

Dendritic spines have been shown to be closely related to various functional properties of the neuron. Usually dendritic spines are manually labeled to analyze their morphological changes, which is very time-consuming and susceptible to operator bias, even with the assistance of computers. To deal with these issues, several methods have been recently proposed to automatically detect and measure the dendritic spines with little human interaction. However, problems such as degraded detection performance for images with larger pixel size (e.g. 0.125 μm/pixel instead of 0.08 μm/pixel) still exist in these methods. Moreover, the shapes of detected spines are also distorted. For example, the "necks" of some spines are missed. Here we present an oriented Markov random field (OMRF) based algorithm which improves spine detection as well as their geometric characterization. We begin with the identification of a region of interest (ROI) containing all the dendrites and spines to be analyzed. For this purpose, we introduce an adaptive procedure for identifying the image background. Next, the OMRF model is discussed within a statistical framework and the segmentation is solved as a maximum a posteriori estimation (MAP) problem, whose optimal solution is found by a knowledge-guided iterative conditional mode (KICM) algorithm. Compared with the existing algorithms, the proposed algorithm not only provides a more accurate representation of the spine shape, but also improves the detection performance by more than 50% with regard to reducing both the misses and false detection.

Neuron tracing in perspective

The study of the structure and function of neuronal cells and networks is of crucial importance in the endeavor to understand how the brain works. A key component in this process is the extraction of neuronal morphology from microscopic imaging data. In the past four decades, many computational methods and tools have been developed for digital reconstruction of neurons from images, with limited success. As witnessed by the growing body of literature on the subject, as well as the organization of challenging competitions in the field, the quest for a robust and fully automated system of more general applicability still continues. The aim of this work, is to contribute by surveying recent developments in the field for anyone interested in taking up the challenge. Relevant aspects discussed in the article include proposed image segmentation methods, quantitative measures of neuronal morphology, currently available software tools for various related purposes, and morphology databases. (c) 2010 International Society for Advancement of Cytometry.

MDL constrained 3-D grayscale skeletonization algorithm for automated extraction of dendrites and spines from fluorescence confocal images

This paper presents a method for improved automatic delineation of dendrites and spines from three-dimensional (3-D) images of neurons acquired by confocal or multi-photon fluorescence microscopy. The core advance presented here is a direct grayscale skeletonization algorithm that is constrained by a structural complexity penalty using the minimum description length (MDL) principle, and additional neuroanatomy-specific constraints. The 3-D skeleton is extracted directly from the grayscale image data, avoiding errors introduced by image binarization. The MDL method achieves a practical tradeoff between the complexity of the skeleton and its coverage of the fluorescence signal. Additional advances include the use of 3-D spline smoothing of dendrites to improve spine detection, and graph-theoretic algorithms to explore and extract the dendritic structure from the grayscale skeleton using an intensity-weighted minimum spanning tree (IW-MST) algorithm. This algorithm was evaluated on 30 datasets organized in 8 groups from multiple laboratories. Spines were detected with false negative rates less than 10% on most datasets (the average is 7.1%), and the average false positive rate was 11.8%. The software is available in open source form.

ONLINE THREE-DIMENSIONAL DENDRITIC SPINES MOPHOLOGICAL CLASSIFICATION BASED ON SEMI-SUPERVISED LEARNING

Recent studies on neuron imaging show that there is a strong relationship between the functional properties of a neuron and its morphology, especially its dendritic spine structures. However, most of the current methods for morphological spine classification only concern features in two-dimensional (2D) space, which consequently decreases the accuracy of dendritic spine analysis. In this paper, we propose a semi-supervised learning (SSL) framework, in which spine phenotypes in three-dimensional (3D) space are considered. With training only on a few pre-classified inputs, the rest of the spines can be identified effectively. We also derived a new scheme using an affinity matrix between features to further improve the accuracy. Our experimental results indicate that a small training dataset is sufficient to classify detected dendritic spines.

Cell segmentation using front vector flow guided active contours

Phase-contrast microscopy is a common approach for studying the dynamics of cell behaviors, such as cell migration. Cell segmentation is the basis of quantitative analysis of the immense cellular images. However, the complicated cell morphological appearance in phase-contrast microscopy images challenges the existing segmentation methods. This paper proposes a new cell segmentation method for cancer cell migration studies using phase-contrast images. Instead of segmenting cells directly based on commonly used low-level features, e.g., intensity and gradient, we first identify the leading protrusions, a high level feature, of cancer cells. Based on the identified cell leading protrusions, we introduce a front vector flow guided active contour, which guides the initial cell boundaries to the real boundaries. The experimental validation on a set of breast cancer cell images shows that the proposed method demonstrates fast, stable, and accurate segmentation for breast cancer cells with wide range of sizes and shapes.