Synopsis

A typical diffusion experiment exists of images acquired at different gradient directions and diffusion weightings. These large datasets necessitate transformation into compact metrics which facilitate comparisons between subjects, brain regions, or experimental settings. This conversion requires the fitting of mathematical functions called signal representations. Signal representations can be motivated by general diffusion physics, microstructural models, or simply by mathematical functions with convenient properties. Depending on the type of signal representation, the fitting parameters can have true physical meaning (e.g., diffusivity), biological meaning (e.g., axon diameter), or they can be a pure mathematical construct (e.g. spherical harmonic coefficients).