The psi_n in the above equation are the normalised frequencies of the transmission zeros.
While the above is fine for getting a filter response on a normalised frequency axis, what you really want for a subsequent synthesis of the transfer function is F(omega) in the form of a polynomial. The above cosh function can indeed be expressed in the form of a polynomial in omega. The recursive process for generating this polynomial is well documented in the 1982 article by Cameron, in his later IEEE MTT articles and in the book by him, Kudsia and Mansour (see book list). My own implementation is available via the link below. A further link shows a screenshot of the program using the very convenient recursive Cameron process.