Ghost imaging is a novel non-conventional technique allowing to generate high resolution images by correlating the intensity of two light beams, neither of which independently contains sufficient information about the spatial distribution and shape of the object. The first demonstration of ghost imaging used light in double photon state, obtained from spontaneous parametric down-conversion. Owing to the entanglement of the source photons, the proposed theory required quantum descriptions for both the optical source and its photo-detection statistics1. However, subsequent experimental and theoretical considerations2,3 demonstrated that ghost imaging can be performed also with thermalized light, utilizing either CCD detector arrays or photon-counting detectors, thus admitting to a semi-classical description, employing classical fields and shot-noise limited detectors. This has generated increasing interest4-6 in establishing a unifying theory that characterizes the fundamental physics of ghost imaging and defines the boundary between classical and quantum domains. In this view, we exploited recent progress obtained through the application of Fourier Transform Techniques to demonstrate ghost imaging in the frequency domain, in order to measure a continuous spectrum by using a highly brilliant and coherent monochromatic source. In particular, we demonstrate the application of this ghost imaging technique to broadband spectroscopic measurements by means of interaction free photon detection. The experimental apparatus and the collected data are described in a dedicated work7. In this paper, we consider the theoretical aspects underlying the proposed Spectroscopic technique. In particular, two alternative theoretical models are presented. In one case, a statistical approach (semi-classical) is applied, where the states of the sampling beam are considered, whereas in the other case a pure quantum treatment is carried on, by describing the interaction of vacuum states generated by photon conversion processes. Both theoretical models, though carried on by means of a complementary formalism, lead to equivalent results and offer a physical interpretation of the collected experimental data. The application of these results offer novel perspectives for remote sensing in low light conditions, or in spectral regions where sensitive detectors are lacking.