NeuroMorph and NeuroCount: automated tools for fast and objective acquisition of neuronal morphology for quantitative structural analysis
-
1
Max-Planck-Institute of Neurobiology, Germany
-
2
Queen Elizabeth The Queen Mother Hospital, United Kingdom
-
3
Konrad-Zuse-Center for Information technology, Germany
The large scale reconstruction of the dendritic and axonal arbor from many neurons and the quantitative determination of neuron distributions across different brain areas are two fundamental tasks in neuroscience. The classification of neuronal cell types, the wiring between these types and simulation of single neuron activity are recent scientific tasks that are based on such anatomical studies. At present most anatomical data is obtained with manual techniques. The standard method to trace neuronal morphology relies on computer aided Camera-Lucida based methods, where a human user performs pattern recognition and marks neuronal structures on a live camera image. The current method to derive neuronal distributions and densities is to manually label cell bodies in confocal image stacks. Both approaches are somewhat subjective, laborious, tedious and hence prone to error. In consequence it is time consuming to obtain a statistically valid sample for quantitative analysis. Currently available automated tracing approaches lack the capability of tracing axons and automated cell counting algorithms usually incorporate an error larger than 10%, making many studies unfeasible. Firstly, we present a novel approach for reconstruction of neuron morphology (NeuroMorph) that is based on transmitted light brightfield mosaic microscopy, high-resolution image restoration and fast, parallel tracing of biocytin labeled neurons. The mosaic technique compensates for the limited field of view and allows the acquisition of high-resolution image stacks on a scale of millimeters. The image restoration by deconvolution is based on experimentally verified assumptions about the optical system. Restoration yields a significant improvement of signal-to-noise ratio and resolution of neuronal structures in the image stack. Application of local threshold and thinning filters result in a 3D graph representation of dendrites and axons in a section. The reconstructed branches are then edited and aligned. Branches from adjacent sections are spliced, resulting in a complete 3D-reconstruction of a neuron. Secondly, we present a novel approach for the automated detection and counting of NeuN stained cell bodies (NeuroCount), which incorporates morphological as well as statically motivated filters. Local filters compensate for inhomogeneity caused by image acquisition or staining. This image stack is further processed by morphological filters resulting in landmarks that represent individual cell bodies. However, the packing density of neurons frequently exceeds the resolution and therefore morphological filters are not capable of splitting all cell clusters. This miscounting is compensated by statistical filters based upon the volume distribution of the labeled objects. The algorithms are not based on any constraints such as shape or size of cell bodies. Therefore the detection performs satisfyingly independent of the method of image acquisition or staining and even if different neuron types are within one image stack. A comparison with manually obtained results showed that our automated approaches are fast and reliable alternatives to the currently available manual systems.
Conference:
Neuroinformatics 2008, Stockholm, Sweden, 7 Sep - 9 Sep, 2008.
Presentation Type:
Poster Presentation
Topic:
Neuroimaging
Citation:
Oberlaender
M,
Broser
P and
Dercksen
V
(2008). NeuroMorph and NeuroCount: automated tools for fast and objective acquisition of neuronal morphology for quantitative structural analysis.
Front. Neuroinform.
Conference Abstract:
Neuroinformatics 2008.
doi: 10.3389/conf.neuro.11.2008.01.065
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
28 Jul 2008;
Published Online:
28 Jul 2008.
*
Correspondence:
Marcel Oberlaender, Max-Planck-Institute of Neurobiology, Munich, Germany, oberlaender@neuro.mpg.de