If you can put your equation in to this form
$\mathbf x^T A \mathbf x = K$ and since A is a symmetric matrix, you can diagonalize it, and not only diagonalize it, diagonalize it with ortho-normal P.
$\mathbf x^T P^T D P\mathbf x = K\\\ (P\mathbf x)^T D P\mathbf x = K$
The matrix $P$ represents a change to the coordinate system -- a basis change. And since $P$ is ortho-normal it is a basis change that is limited to reflections and rotations i.e. distance and volumes are preserved. It doesn't change the nature of the conic section.
$\mathbf y = P\mathbf x$ and $\mathbf y^T D \mathbf y = K$