Hyperspectral cathodoluminescence imaging provides spectrally and spatially resolved information on luminescent materials within a single dataset. Pushing the technique toward its ultimate nanoscale spatial limit, while at the same time spectrally dispersing the collected light before detection, increases the challenge of generating low-noise images. This article describes aspects of the instrumentation, and in particular data treatment methods, which address this problem. The methods are demonstrated by applying them to the analysis of nanoscale defect features and fabricated nanostructures in III-nitride-based materials.