In order to study spray combustion, an experimental test rig was developed at ONERA to partially characterize the flow conditions inside the combustion chamber of a gas turbine. Experimental campaigns using laser-based diagnostics were performed to provide an experimental database under reacting and non-reacting conditions. The paper first describes the Mie scattering image-processing to detect the droplets in the spray, and to calculate 2D maps of droplet number density and mean inter-droplet distance. The method is subsequently used to investigate the spray behavior under both reacting and non-reacting conditions according to global-averaging and phase-averaging methods. Experimental findings on the spatial droplet distribution in the spray are compared to the simple regular grid distribution and the Hertz-Chandrasekhar distribution. Results show that, under both conditions, there is an affine relationship between the inverse square root of the mean droplet number density and the nearest-neighbor inter-droplet distance. Moreover, observations suggest that the droplet spatial distribution fits more closely to a Hertz-Chandrasekhar distribution than a simple regular grid distribution, which may bring new insight for spray modeling.