Quantitative microwave holography is a recent imaging methodology that shows promise in medical diagnostics. It is a real-time direct inversion algorithm that reconstructs the complex permittivity from S-parameter measurements on an acquisition surface outside of the imaged object. It is recognized that this imaging method suffers from limitations in tissue imaging due to a forward model which linearizes a highly nonlinear scattering problem. It is therefore important to study its limitations when reconstruction is aided by certain pre-and post-processing filters which are known to improve the image quality. The impact of filtering on the quantitative result is particularly important. In this work, the reconstruction equations of quantitative microwave holography are derived from first principles. The implementation of two linearizations strategies, Born's approximation and Rytov's approximation, is explained in detail in the case of a scattering model formulated in terms of S-parameters. Furthermore, real-space and Fourier-space filters are developed to achieve the best performance for the given linearized model of scattering. Simulated and experimental results demonstrate the limitations of the method and the necessity of filtering. The two approximations are also compared in experimental tissue reconstructions.