Nearest Neighbor Density Analysis of the Stellate Ganglia

Abstract

IntroductionThe so‐called heart‐brain axis is the neurological and physiological connection between the autonomic systems of both organs. The stellate ganglia and superior cervical ganglia are the location of the sympathetic neuronal cells that innervate the vasculature of the heart and brain, respectively. The paravertebral chain connects the ganglia on either side of the neck. Myocardial infarction is thought to cause remodeling of the sympathetic neurons in the stellate ganglia. In a recent qualitative study, the stellate ganglia were shown to remodel with regard to cell body number, size, density, and clustering in the presence of various cardiac diseases. We developed a tool to automatically quantify the nearest neighbor density of neurons weighted by size in pathology specimens of the stellate ganglia. This is the first step to quantifying sympathetic nervous system changes in the heart‐brain axis of cardiovascular pathology.MethodsBiopsies of the stellate ganglia were collected bilaterally from an opportunity sample of 17 willed body donors (34 total specimens) with and without histories of self‐reported myocardial infarction (MI) and/or cerebrovascular accident (CVA). The ganglia were stained with luxol fast blue and cresyl violet which allowed for histological analysis of local neurons. Images of the ganglia at its middle coronal slice after staining were processed using multiple libraries in python, primarily PIL and OpenCV. The images first underwent multiple filters: a high contrast filter and a smoothing filter. After this filter, the image passed through a color threshold that obtained the location of each neuron; the locations of each neuron were used to make a binary image. The binary image was passed through various parameters such as neuron area, radius of the minimum enclosing circle, and a ratio of the area of the neuron and area of the minimum enclosing circle. A heatmap was generated from this last image through a convolution calculation that takes into account the location of the neuron, size, and proximity of surrounding neurons‐ all variables that are needed to calculate nearest neighbor density. Quantitative and qualitative measurements are generated from this: a graphical map of high and low density areas and number of neurons per given threshold of density.ResultsA pipeline was created that generates heatmaps of the nearest neighbor density based on the location and size of neurons on the stellate ganglia. Quantitative data is also measured based on how many neurons are located in varying levels of density structures.ConclusionsHeatmaps of the nearest neighbor density and size of the neurons potentially show signs of cardiac diseases‐ regions of high density or patterns of where highly dense regions are located are indicators. Developments within computation pathology is an easily accessible and pioneering way for new technological innovation.