my understanding is that the op is wants to use fourier methods (i.e. orthogonal functions) to develop a solution for an elliptical (presumed 2d) region.
Carslaw and Yeager might be a good start. the expansions are not limited to trig expansions and are more generally classified as integral methods.