Microwave imaging has many applications in the biomedical area, and one of the most promising application is brain stroke detection. The full-wave 3D modeling of medical imaging scenarios requires significant computational resources due to the complex anthropomorphic phantoms and antenna systems. Hence, it is of paramount interest to simplify unnecessary details in numerical models, without sacrificing accuracy. Algorithms for reconstruction are based on scattering parameters and electric field in the domain of interest. Hence, by comparing the S-parameters of the original model and the simplified one, we can assess the quality of the simplification. Here, we propose the algorithm for computing the equivalent homogeneous phantom from a realistic human head model. In the qualitative algorithms, the equivalent phantom can be used for the reference model as the patient's head before the stroke onset. In quantitative algorithms, such model can be employed as the intelligent solution for initializing the iterative process.