A software package for
Automated Neuronal Morphology

W. B. Lindquist and C.M. Weaver

Software Authors:   Prof. W. B. Lindquist,   Dr. I.Y.Y. Koh,   Dr. C.M. Weaver

Package Overview (this document)
System Requirements
Source Code Availability
Distribution Installation and Compiling Instructions
Instructions for Running 3DMA-Neuron
Release Notes and Bug Reports

Package Overview

Table of Contents

I) Introduction

II) Morphology of Dendritic Spines

1) Segmentation
2) Construction and Modification of the Medial Axis
3) Dendritic Backbone Extraction
4) Spine Detection
5) Manual Editing

III) Dendritic Branching Morphology

1) Segmentation
2) Obtaining the Neural Skeleton
3) Branch Point Labeling
4) Morphological Measurements

IV) Multiscale Morpholgy

V) References

VI) Funding Acknowledgements

I) Introduction

3DMA-Neuron developed as an extension of the 3DMA software package for analyzing biphase two- and three-dimensional images.  3DMA-Neuron has the capability to perform automated, detailed analyses of 3D optical images containing dendritic branch segments showing dendritic spines [7], which we refer to as fine structure analysis.. It also has the capability to perform automated branching morphology on the dendritic trees of 3D images of single neurons [8,10]. Further, the dendrite and spine morphology functionalities have been integrated, so that both the branching morphology as well as the fine scale structure of a neuron can be analyzed when a neuron is imaged at sufficiently high resolution.   The high degree of automation and relative speed in computational time make 3DMA-Neuron well suited to obtaining detailed morphology of multiple 3D images, allowing comparison among different samples.

The image analysis algorithms (for both the dendritic spine detection and the dendritic branching detection) consist of three general steps:  image segmentation, extraction of the medial axis of the neuron phase, and analysis of the medial axis and neuron phase.  These steps are discussed below, largely by illustration, for both applications.  Also illustrated are the morphological measurements obtainable after the appropriate dendritic structures are identified.

This figures accompanying this html file are best viewed with the html viewer window set to full screen.

II) Morphology of Dendritic Spines

The principal components of the fine structure analysis consist of spine detection, morphological characterization of each spine, and dendritic radius determination. The algorithms used for spine detection/characterization have been described in detail elsewhere [7] and have four main components:  segmentation; dendritic backbone extraction;  spine detection and merging; and measurement of morphological characterizations.

II.1) Segmentation

The greyscale intensity of each voxel is an integer from 0-255 (for 8-bit images). 3DMA-Neuron provides two options for image segmentation:  simple segmentation and indicator kriging.  For simple segmentation (Fig. 4), the user must enter a single threshold:  voxels having intensity above this threshold are assigned to the neuron phase; remaining voxels are assigned to the background phase.  Indicator kriging [12] is a locally adaptive segmentation method, which requires that a subpopulation of voxels of each phase be positively identified a priori.  This can be done by manually establishing a window of intensity values delimited by two thresholds. Intensity values above the upper threshold are assumed neuron, those below the lower threshold are assumed background tissue. The classification of remaining voxels is estimated by indicator kriging which utilizes an estimate of a correlation function incorporating local spatial information.  Fig. 1 shows a image segmented by indicator kriging.

Fig. 1: xy-projections of (left) a laser scanned, confocal microscope image of part of the dendritic tree from a pyramidal neuron and (right) the neuron phase (shown in black) identified after segmentation by the indicator kriging method.

II.2) Construction and Modification of the Medial Axis

The medial axis of a digitized object is the minimally connected skeleton by which the original topology and geometry of the object is preserved.  The medial axis of the neuron is composed of voxels from the neuron phase which are equidistant from voxels on the neuron surface.  Erosion-based algorithms are used to construct the medial axis; 3DMA-Neuron employs the algorithm developed by Lee et al. [9].  The medial axis is sensitive to surface noise; as a result, the medial axis may contain spurious paths that do not contribute to the overall geometry of the object.  Such spurious paths are trimmed using a minimum length criterion.  Fig. 2 shows the medial axis extracted from the raw data shown in Fig. 1(left), before and after spurious paths are trimmed.

Fig. 2:  Shown are xy-projections of (left) the medial axis extracted from the neuron phase identified in Fig.1(right); (right) the medial axis, after trimming short spurs and resolving loops.  The colors represent, in rainbow scale, the diameter of the dendrite at each medial axis voxel.

For more information on the medial axis algorithms, see the Online Manual to 3DMA.

II.3) Dendritic Backbone Extraction

After the medial axis has been constructed and trimmed, the dendritic backbones are extracted from the medial axis by trimming short `spurs' and resolving loops.  The user may select whether a dendritic backbone will be continued through a branch point.  If the option to terminate backbones at branch points is not selected, a backbone is continued through a branch point according to a minimum deviation angle criterion as described in [7].

II.4) Spine Detection

Spine detection proceeds in three steps: detection of spine components entirely detached from dendrites containing backbones, detection of spine components attached to these same dendrites, and merging of two or more spine components.  Any dendritic phase component containing no dendritic backbone whose center of mass is sufficiently close to a dendritic backbone is identified as a detached spine head.  Attached spines are detected as local protrusions of a dendritic surface.  To identify spines which comprise more than one attached or detached spine component, a final merging algorithm is performed.  Fig. 3 shows the result of spine detection on the raw data shown in Fig. 1.

Fig. 3:  Shown are xy-projections of dendritic backbones and spines identified from Fig. 1 after the spine detection algorithms are completed.  Barely visible in this figure is the backbone of each identified dendritic segment.  The color scheme is arbitrary.

II.5) Manual Spine Editing

As a final level of user control, the 3DMA-Neuron graphical user interface (GUI), provides a facility by which the user can scan projected images showing identified spines, and by mouse click remove spines (or dendritic segments) deemed to be false positive signals.

III) Dendritic Branching Morphology

For a typical neuron image, there are three main components to the algorithms we employ for determination of the dendritic branching morphology: segmentation of the raw data to distinguish the neuron phase from the background phase; identification of the dendritic structure using skeletonization; and labeling of branch points (and hence branch segments) with respect to branch order.  These are described in detail elsewhere [8].

III.1) Segmentation

See section II.1 for a description of the available segmentation algorithms. In Fig. 4, results from simple segmentation of full neuron image are illustrated. In contrast to the kriging results shown in Fig. 1, note the speculation "noise" in the background tissue to which simple segmentation is prone.


Fig. 4:  xy-, xz-, and yz-projections of (left) raw and (right) simply segmented data of a pyramidal neuron. 

III.2) Obtaining the Neural Skeleton

The largest connected component of the neuron phase is retained as comprising the cell; disconnected components are discarded as debris.  Once this is done, both the medial axis extraction and trimming are performed as described above in the spine detection application.  Fig. 5 shows the trimmed medial axis result.

Fig. 5:  xy-projection of the extracted skeleton of the neuron pictured in Fig. 4.

III.3) Branch Point Labeling

The limits of the soma are delineated manually using the GUI [8]. The skeleton inside the soma region is omitted from further analysis.  We refer to each dendritic trunk leaving the soma as the root of a dendritic tree.  Using the GUI, manual point and click is used to select one (or more) trees that will be identified as the apical arbor.  All other trees comprise the basal arbor.  Any axon collaterals which are detected can be removed manually.  Dendritic branches in each tree are labeled according to a centrifugal nomenclature [14].  Looping structures in the skeleton are resolved prior to labeling.  The final result is shown in Fig. 6 for the apical and basal brushes separately.


Fig. 6:  (a) Apical and (b) basal arbors detected, after manual selection of the soma region and apical brush.  The skeleton segments designated as interior to the soma region are shown in pink.  Remaining colors, in rainbow scale, correspond to the branch orders of the neuron branch segments.

III.4) Morphological Measurements

Recorded for each spine are:  length, volume, head and neck diameters, position on the dendritic backbone, and shape classification (thin, mushroom, stubby) [11, 5].  Spines which appear disconnected are assigned a neck diameter of 0 and are designated as mushroom shaped.  Spine volume is determined to be sum of the volume of all voxels contained in the spine.

Recorded for each dendritic branch are:  total length, running diameter, number of spines, spine density, mean spine length, and relative populations of spines of each classification type.  Dendritic diameter determination arises directly from the skeletonization procedure used to determine the backbone of each dendritic branch segment lying in each tile. The radius is measured as a function of position along any dendritic branch segment.

The branching morphology is output in SWC format [2], a standard text file which lists basic information about the soma and each dendritic branch.  More detailed output, including summary details about the spine shapes on each branch, can be output in NeuroML format [4, 3], a markup language which serves as an interchange format between various software tools.  These formats allow the data to be easily incorporated into compartmental modeling software such as NEURON [6] and GENESIS [1].

IV) Multiscale Morphology

We have integrated the dendritic branching (section III) and spine detection (section II) algorithms to obtain multiscale morphological analysis of pyramidal neurons.  An entire pyramidal neuron (Fig. 7) from the macaque monkey superior temporal cortex has been imaged in three dimensions at high resolution (0.098 x 0.098 x 0.081 um) via confocal laser scanning microscopy, in several stacks.  The various stacks were integrated into a single volume as described in [13], resulting in a 25 GB data set.  Such a data set is too large to be analyzed as a single volume by the computational power accessible to most labs.

Fig. 7:  xy-projection of an entire pyramidal neuron from the macaque monkey superior temporal cortex, imaged at 0.098 x 0.098 x 0.081 um resolution.  Individual tiles manually selected for spine detection analysis are also indicated.

For dendritic spine morphology analysis, the data set was first subdivided into non-overlapping tiles; algorithms to detect spine characteristics were then applied to each tile (see Section II).  To analyze the dendritic branching morphology, the global coordinates of all detected dendrite fragments are mapped into a smaller cube.  This provides a representation of the dendritic structure amenable to automated branching morphology analysis (section III ).  An alignment procedure was applied to associate the spine data from each fine scale dendritic fragment to the corresponding neuron branch identified in the compressed structure.  The result is a detailed morphometric description of the entire neuron (Fig. 8).  For more detail on these algorithms, see [16].




Figure 8:  Reconstructed (a) apical and (b) basal arbors of the neuron in Fig. 7.  In (a), the apical shaft (shown in black), the terminal tuft (shown in red), and oblique branch dendrites (shown in remaining colors) are identified.  (c) shows a close-up of some of the branches.  The dots represent spines associated with the dendrites.

V) References

[1]    J. M. Bower and D. Beeman.  The Book of GENESIS:  Exploring realistic neural models with the General Neural Simulation System.  TELOS/Springer-Verlag, 1994.

[2]    R. C. Cannon, D. A. Turner, G. K. Pyapali, and H. V. Wheal.  An on-line archive of reconstructed hippocampal neurons.  J. Neurosci. Methods, 84:  49-54, 1998.

[3]    N. H. Goddard, D. Beeman, R. Cannon, H. Cornelis, M. O. Gewaltig, G. Hood, F. Howell, P. Rogister, E. De Schutter, K. Shankar, and M. Hucka.  NeuroML for plug and play neuronal modeling.  Neurocomputation 44:  1077-1081, 2002.

[4]    N. H. Goddard, M. Hucka, F. Howell, H. Cornelis, K. Shankar, and D. Beeman.  Towards NeuroML:  model description methods for collaborative modeling in neuroscience. Philos. Trans. R. Soc. B, 356:  1209-1228, 2001.

[5]    K. M. Harris, F. E. Jensen, and B. Tsao.  Three-dimensional structure of dendritic spines and synapses in rat hippocampus (CA1) at postnatal day 15 and adult ages:  implications for the maturation of synaptic physiology and long-term potentiation. J. Neurosci. 12:  2685-2705, 1992.

[6]    M. L. Hines.  The Neuron simulation program.  In J. Skrzypek, editor, Neural Network Simulation Environments, pages 147-163.  Kluwer, Norwell, MA, 1994.

[7]    I. Y. Y. Koh, W. B. Lindquist, K. Zito, E. A. Nimchinsky, and K. Svoboda.  An image analysis algorithm for dendritic spines.  Neural Comput., 14:  1283-1310, 2002.

[8]    Y. Y. Koh.  Automated recognition algorithms for neural studies.  PhD thesis, Stony Brook University, May 2001.

[9]    T. C. Lee, R. L. Kashyap, and C. N. Chu.  Building skeleton models via {3-D} medial surface/axis thinning algorithms.  CVGIP:  Graph. Models Image Process. 56:  462-478, 1994.

[10]    M. Maravall, G. M. G. Shepherd, Y. Y. Koh, W. B. Lindquist, and K. Svoboda.  Experience-dependent changes in dendritic morphology of layer 2/3 pyramidal neurons during a critical period for developmental plasticity in rat barrel cortex.  Submitted to Cerebral Cortex.

[11]    E. A. Nimchinsky, B. L. Sabatini, and K. Svoboda.  Structure and function of dendritic spines. Ann. Rev. Physiol. 64:  313-352, 2002.

[12]    W. Oh and W. B. Lindquist, Image thresholding by indicator kriging. IEEE Trans. Pattern Anal. Mach. Intell. 21:  590-602, 1999.

[13]    A. Rodriguez, D. Ehlenberger, K. Kelliher, M. Einstein, S. C. Henderson, J. H. Morrison, P. R. Hof, and S. L. Wearne.  Automated reconstruction of three-dimensional neuronal morphology from laser scanning microscopy images.  Methods 30:  94-105, 2003.

[14]    H. B. M. Uylings, A. Ruiz-Marcos, and J. van Pelt.  The metric analysis of three-dimensional dendritic tree patterns:  a methodological review. J. Neurosci. Methods 18:  127-151, 1986.

[15]    C. M. Weaver.  Automated morphometry for neural cells.  PhD thesis, Stony Brook University, August 2003.

[16]    C. M. Weaver, W. B. Lindquist, S. L. Wearne, and P. R. Hof.  Automated algorithms for multiscale morphometry of neural cells.  In preparation, 2003.

VI) Funding Acknowledgements
  Support for the development of this software package was provided, in part, by the National Science Foundation, Grant #0107893. Any opinions, findings and conclusions or recomendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF).