Graphic User Interface for Reliable and Repeatable Quantification of Neuron Morphology and Microstructural Analysis

Abstract

Steroid hormones influence anxiety-like behavior and interact with the prototypical anti-anxiety medication, diazepam (Valium) in humans and in animals [1]. In Long Evans female rats, the proestrus stage of the estrous cycle (when estradiol and progesterone levels peak) increases anxiety-like behavior as well as improvements in diazepam efficacy [2]. Progesterone and diazepam modulate the γ-aminobutyric acid (GABA) system, which is implicated in anxiety disorders and important in neural development. Thus GABA plays a major inhibitory role in mammalian cortex. Additionally, estrous changes also relate to fluctuations in brain-derived neurotrophin factor (BDNF), with BDNF expression or efflux peaking and falling with rising estradiol and declining progesterone, respectively. Fluctuations in BDNF are also implicated in anxiety-like behavior in animals and the pathogenesis and treatment of anxiety disorders in humans [3]. Given this previous hormonal, pharmacokinetic and morphological data implicating elevated responses to amphetamine and more pronounced stimulant sensitization in females, there is a need to examine any acute changes in terms of stained neuron count, size, eccentricity, aspect ratio, and count density in medial prefrontal cortex GABAergic neuron levels.To carry out quantitative analysis, it is common practice to perform the analysis manually because the notion of determining what is and is not significant in images of new research (and therefore inherently low data sets) has yet to be algorithmically solved. This process not only introduces systematic and hard-to-characterize biases, but it is extremely time intensive. Through interdisciplinary collaboration, we developed an open source user-friendly graphic user interface (GUI) in MATLAB from the ground up shown in figure 1, for neuron reconstruction and microstructural analysis that addresses unmet demands posed by current problems in biomedical research-capturing dynamic changes in neurons and objectively quantifying morphology differences at a cost-effective level. Fixed brain post-mortem coronal cross-sections were mounted on to motorized Zeiss AxioObserver inverted microscope that is equipped with a CCD camera (AxioCam HRm3) and connected to a desktop computer, to capture tiled images in a Z-stack. User selected brain regions are then imaged using a 40x objective (EC Plan-Neofluar 40x/0.75) in a tiling pattern with 5% overlap. Each brain region was then focus stacked, stitched in a batch process via ImageJ macros, and fed into the custom GUI developed in MATLAB, as shown in figure 1. Within the GUI, we stitch images using a powerful, optimizes version of phase correction [4]. Otsu's method and maximum entropy thresholding are used to separate the background and physical thresholds such as minimum and maximum size, eccentricity, circularity of neuron are used to determine what is and is not significant in the image. Resultant images are then used to quantify BDNF and PV-positive immunoreactive images in high and low anxiety rats. We counted imm...