Quantifying How Staining Methods Bias Measurements of Neuron Morphologies

Abstract

The process through which neurons are labeled is a key methodological choice in measuring neuron morphology. However, little is known about how this choice may bias measurements. To quantify this bias we compare the extracted morphology of neurons collected from the same rodent species, experimental condition, gender distribution, age distribution, brain region and putative cell type, but obtained with 19 distinct staining methods. We found strong biases on measured features of morphology. These were largest in features related to the coverage of the dendritic tree (e.g., the total dendritic tree length). Understanding measurement biases is crucial for interpreting morphological data.

Keywords: dendritic morphology; fluorescence microscopy; golgi method; immunostaining; neuroinformatics; rodent neuroanatomy; staining method.

Figure 1

The spatial distribution of the 22 groups analyzed are from multiple brain regions in rodents. The number of comparison groups in each region are listed in parentheses.

Figure 2

Sample neuron morphologies and features. (A) Sample morphologies from two groups of neurons, comparing two distinct staining methods. (B) Geometrical features of neuron morphology. Local angle represents the angle between adjacent edges not at a branch point. The global angle represents the angle between an edge and the vector pointing toward the soma. The branch angle represents the angle between two edges that branch from a common node. (C) Histograms for the two sample groups of the six morphological features used in analysis.

Figure 3

Proportion of significant differences between pairs of staining methods over all groups and morphological features. Zero means there is not any difference between the staining methods among the groups and one means all the groups are different. Computed using the Wilcoxon rank-sum test, corrected (Equation 2). Significant means p < 0.05, highly significant means p < 0.001.

Figure 4

Pairwise average effect sizes for six morphological features. Upper right entries show statistical significance of differences (Equation 3). White squares represent no comparison, light gray squares represent a non-significant difference, gray squares represent a significant difference (p < 0.05), and black squares represent a highly significant difference (p < 0.001). To compute the significance level the average absolute difference in morphological features is compared with a null-distribution generated through permutation. Lower left entries show average difference in each feature between the two corresponding methods.