Semi-automatic quantification of neurite fasciculation in high-density neurite images by the neurite directional distribution analysis (NDDA)

Abstract

Background: Bundling of neurite extensions occur during nerve development and regeneration. Understanding the factors that drive neurite bundling is important for designing biomaterials for nerve regeneration toward the innervation target and preventing nociceptive collateral sprouting. High-density neuron cultures including dorsal root ganglia explants are employed for in vitro screening of biomaterials designed to control directional outgrowth. Although some semi-automated image processing methods exist for quantification of neurite outgrowth, methods to quantify axonal fasciculation in terms of direction of neurite outgrowth are lacking.

New method: This work presents a semi-automated program to analyze micrographs of high-density neurites; the program aims to quantify axonal fasciculation by determining the orientational distribution function of the tangent vectors of the neurites and calculating its Fourier series coefficients ('c' values).

Results: We found that neurite directional distribution analysis (NDDA) of fasciculated neurites yielded 'c' values of ≥∼0.25 whereas branched outgrowth led to statistically significant lesser values of <∼0.2. The 'c' values correlated directly to the width of neurite bundles and indirectly to the number of branching points.

Comparison with existing methods: Information about the directional distribution of outgrowth is lost in simple counting methods or achieved laboriously through manual analysis. The NDDA supplements previous quantitative analyses of axonal bundling using a vector-based approach that captures new information about the directionality of outgrowth.

Conclusion: The NDDA is a valuable addition to open source image processing tools available to biomedical researchers offering a robust, precise approach to quantification of imaged features important in tissue development, disease, and repair.

Keywords: Axon branching; Axonal alignment; Axonal bundling; Dorsal root ganglia; Fasciculation; Image processing; Neurite quantification methods.

Figure 1

Image processing for NDDA for quantification of fasciculation. (a) Fluorescent microscopy images (4x) of DRG explant outgrowth on silk hydrogels before (i) and after (ii) artifact removal with Adobe Photoshop. Scale bars are omitted during processing to prevent interference with analysis program. (b–d) Image processing steps within Mathematica program. (b) Background removal and binary conversion. (c) Tangent vector assignment. (d) Local radial frame definition to eliminate floating neurites. Neurites immunostained with α-β-III tubulin. Scale bars = 150μm.

Figure 2

NDDA output and ‘c’ values for quantification of fasciculation. (a) Histogram of overall DRG outgrowth summarizing counts of tangent vectors from ϕ1 to ϕ2. (b) Spatial resolution of fasciculation analysis. Histograms (left) of corresponding highlighted annulus (right). Greater numbers of tangent vectors in alignment with the radial direction (zero angular separation) correspond to greater fasciculation. Scale bars = 150μm.(c) Quantification of fasciculation of overall outgrowth as well as selected outgrowth shown in (b). Higher ‘c’ values correspond to greater fasciculation. Scale bars are omitted during processing due to interference with analysis program.

Figure 3

Quantification of fasciculation based on NDDA ‘c’ values. (a) Total ‘c’ values of branched as compared to fasciculated samples. n≥10, error = SEM, *p=0.0171 (t-test). (b) Representative output histograms and images of NDDA of branched (i) and fasciculated (ii) samples. Scale bars omitted due to interference with analysis program. Scale bars = 150μm. (c) Spatially resolved NDDA fasciculation quantification. c1-4 represent annuli with increasing distance from explant body as displayed in (b). Error = SEM, *p = 0.0385 (t-tests), ***p=0.0004 (2-way ANOVA).

Figure 4

NDDA method supplements the Sholl analysis and manual ImageJ neurite width calculations. (a) Immunofluorescent images (neuronal specific β-III Tubulin) of highly branched (i) and fasciculated (ii) DRG outgrowth. Scale bars = 100μm. (b) Quantification of # of intersections and ODF ‘c’ values of images in (a) by the Sholl analysis (i) and NDDA (ii), respectively. (c) Comparison of NDDA ‘c’ values with manual measurements of neurite width. **p=0.0036, ***p<0.0001 (Student’s t-test).