Learning-based long-range axon tracing in dense scenes

Hernandez, Mark; Brewster, Adam; Thul, Larry; Telfer, Brian A.; Majumdar, Arjun; Choi, Heejin; Ku, Taeyun; Chung, Kwanghun; Brattain, Laura J.

doi:10.1109/isbi.2018.8363875

Cited by 5 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Several automated/semi-automated algorithms for tracing and/or segmentation of individual neurons have had some success for certain types of data (Santamaria-Pang et al, 2015;Zhou et al, 2015;Quan et al, 2016;Hernandez et al, 2018;Callara et al, 2020). For the non-uniform background noise problem, the "smart region-growing algorithm" (SmRG) (Callara et al, 2020) segments the neurons using local thresholding based on the distributions of foreground and background signals of optical microscopy.…”

Section: Discussionmentioning

confidence: 99%

NeuroRetriever: Automatic Neuron Segmentation for Connectome Assembly

Shih

Chen

Wang

et al. 2021

Front. Syst. Neurosci.

View full text Add to dashboard Cite

Segmenting individual neurons from a large number of noisy raw images is the first step in building a comprehensive map of neuron-to-neuron connections for predicting information flow in the brain. Thousands of fluorescence-labeled brain neurons have been imaged. However, mapping a complete connectome remains challenging because imaged neurons are often entangled and manual segmentation of a large population of single neurons is laborious and prone to bias. In this study, we report an automatic algorithm, NeuroRetriever, for unbiased large-scale segmentation of confocal fluorescence images of single neurons in the adult Drosophila brain. NeuroRetriever uses a high-dynamic-range thresholding method to segment three-dimensional morphology of single neurons based on branch-specific structural features. Applying NeuroRetriever to automatically segment single neurons in 22,037 raw brain images, we successfully retrieved 28,125 individual neurons validated by human segmentation. Thus, automated NeuroRetriever will greatly accelerate 3D reconstruction of the single neurons for constructing the complete connectomes.

show abstract

Section: Discussionmentioning

confidence: 99%

NeuroRetriever: Automatic Neuron Segmentation for Connectome Assembly

Shih

Chen

Wang

et al. 2021

Front. Syst. Neurosci.

View full text Add to dashboard Cite

show abstract

“…Indeed, many SoA algorithms perform extraordinarily well with low-quality images possessing noisy points, large gaps between neurites and non-smooth surfaces (Liu et al, 2016), since they were likely developed specifically for such purposes. On the contrary, they may perform modestly or even fail in reconstructing densely-packed neurons (Hernandez et al, 2018), such as PCs in the murine cerebella because the Coefficients of variation for neuron volume, area and AUC for n different RG seeds. *The coefficient of variation corresponds to the standard deviation divided by the mean σ/µ.…”

Section: Discussionmentioning

confidence: 99%

“…However, building the training dataset is very time consuming, in particular because it needs to be fleshed out when dealing with different images (e.g., neuron types with different morphology or stacks with different background/foreground features). Finally, many tools and algorithms for neuron segmentation primarily focus on sparsely labeled data, such that their application to images (or volumes) representing densely packed neurons, typical of mammalian brains, is limited (Chothani et al, 2011;Wang et al, 2011Wang et al, , 2017Peng et al, 2014;Hernandez et al, 2018).…”

Section: Introductionmentioning

confidence: 99%

A Smart Region-Growing Algorithm for Single-Neuron Segmentation From Confocal and 2-Photon Datasets

Callara

Magliaro

Ahluwalia

et al. 2020

Front. Neuroinform.

View full text Add to dashboard Cite

Accurately digitizing the brain at the micro-scale is crucial for investigating brain structure-function relationships and documenting morphological alterations due to neuropathies. Here we present a new Smart Region Growing algorithm (SmRG) for the segmentation of single neurons in their intricate 3D arrangement within the brain. Its Region Growing procedure is based on a homogeneity predicate determined by describing the pixel intensity statistics of confocal acquisitions with a mixture model, enabling an accurate reconstruction of complex 3D cellular structures from highresolution images of neural tissue. The algorithm's outcome is a 3D matrix of logical values identifying the voxels belonging to the segmented structure, thus providing additional useful volumetric information on neurons. To highlight the algorithm's full potential, we compared its performance in terms of accuracy, reproducibility, precision and robustness of 3D neuron reconstructions based on microscopic data from different brain locations and imaging protocols against both manual and state-of-the-art reconstruction tools.

show abstract

“…From a historical perspective, the majority of the algorithms performing single neuron reconstructions are primarily focused on sparsely labeled data acquired from the nervous systems of insects or worms (Wang et al, 2011; Quan et al, 2016). Thus their application to this new class of images, usually representing dense-packed neurons typical of mammalian brains, is limited (Chothani et al, 2011; Wang et al, 2017; Hernandez et al, 2018). For this reason, manual segmentation is still considered the gold-standard.…”

Section: Brief Historical Perspectivementioning

confidence: 99%

Gotta Trace ‘em All: A Mini-Review on Tools and Procedures for Segmenting Single Neurons Toward Deciphering the Structural Connectome

Magliaro

Callara

Vanello

et al. 2019

Front. Bioeng. Biotechnol.

View full text Add to dashboard Cite

Decoding the morphology and physical connections of all the neurons populating a brain is necessary for predicting and studying the relationships between its form and function, as well as for documenting structural abnormalities in neuropathies. Digitizing a complete and high-fidelity map of the mammalian brain at the micro-scale will allow neuroscientists to understand disease, consciousness, and ultimately what it is that makes us humans. The critical obstacle for reaching this goal is the lack of robust and accurate tools able to deal with 3D datasets representing dense-packed cells in their native arrangement within the brain. This obliges neuroscientist to manually identify the neurons populating an acquired digital image stack, a notably time-consuming procedure prone to human bias. Here we review the automatic and semi-automatic algorithms and software for neuron segmentation available in the literature, as well as the metrics purposely designed for their validation, highlighting their strengths and limitations. In this direction, we also briefly introduce the recent advances in tissue clarification that enable significant improvements in both optical access of neural tissue and image stack quality, and which could enable more efficient segmentation approaches. Finally, we discuss new methods and tools for processing tissues and acquiring images at sub-cellular scales, which will require new robust algorithms for identifying neurons and their sub-structures (e.g., spines, thin neurites). This will lead to a more detailed structural map of the brain, taking twenty-first century cellular neuroscience to the next level, i.e., the Structural Connectome.

show abstract

Learning-based long-range axon tracing in dense scenes

Cited by 5 publications

References 12 publications

NeuroRetriever: Automatic Neuron Segmentation for Connectome Assembly

NeuroRetriever: Automatic Neuron Segmentation for Connectome Assembly

A Smart Region-Growing Algorithm for Single-Neuron Segmentation From Confocal and 2-Photon Datasets

Gotta Trace ‘em All: A Mini-Review on Tools and Procedures for Segmenting Single Neurons Toward Deciphering the Structural Connectome

Contact Info

Product

Resources

About