ABSTRACT
This study describes the measurements of dielectric properties in the microwave range to differentiate various human central nervous structures. Using a vector network analyser transmission and reflection coefficients were measured from 500 MHz to 18 GHz in four human formalin fixed human brains. The positions of the electrodes were marked, and the tissue was histologically stained to visualize the myelo- and the cytoarchitecture as well as the nerve fibre orientation at the electrodes. The profiles of the transmission coefficients showed a characteristic minimum peak. In order to describe this peak, a mathematical function was fitted. Parameters derived from digital image processing were used to characterize the myelo- and cytoarchitecure of the tissue at the electrodes. A multiple regression model, with the frequency at the transmission peak minimum as a dependent variable and two tissue characteristics at the two electrodes as independent variables, showed a multiple regression coefficient of 0.765. A neural network model was able to estimate the frequency at the transmission peak minimum from the tissue characteristics at the electrode. The measurements of dielectric properties are well suited to differentiate distinct intracerebral structures. The method could be used for online monitoring of the needle's position during a stereotactic intervention in neurosurgery.