Confocal Microwave Imaging for Brain Diagnostics: A Performance Assessment of DAS and MVDR Beamforming Algorithms using Realistic Head Models

Confocal Microwave Imaging (CMI) has recently emerged as a promising non-invasive method for detecting brain tumors, leveraging the dielectric difference between healthy and diseased tissues. This study evaluates the effectiveness of two leading beamforming algorithms—Delay and Sum (DAS) and Minimum Variance Distortionless Response (MVDR)—for confocal image reconstruction using simulated S-parameter data from CST Studio Suite v2024. Various head models were used, including homogeneous, symmetric heterogeneous, and realistically detailed heterogeneous (HUGO voxel) models. Both single- and multiple-tumor scenarios were explored. The MVDR algorithm significantly outperformed DAS in terms of tumor localization and spatial resolution, particularly in complex models and deep tissue layers, as demonstrated in our simulations. Adaptive beamforming techniques also achieved better spatial accuracy and contrast in heterogeneous brain phantoms. These results highlight the diagnostic potential of MVDR-based CMI systems and lay the groundwork for future advancements through the use of larger antenna arrays and the integration of machine learning strategies to enhance imaging performance.