Multistatic Biomedical Microwave Imaging Using Spatial Interpolator for Extended Virtual Antenna Array

The accuracy of multistatic microwave imaging is highly dependent on the number of antennas used for data acquisition. The antenna size, available space for antennas, mutual coupling between antennas and acceptable hardware complexity (switching and processing) limit the usable number of antennas. To address this issue, the concept of virtual array is utilized. In this regard, a spatial interpolator is designed to predict the received signals at the location of the virtual elements using the recorded signals by a limited number of real antennas. Consequently, a frequency-based imaging algorithm is used to process the virtual-array signals and produce clear images that enable accurate detection. The presented method is tested via simulations and experiments using a multistatic-radar-based head imaging system operating using the band 1.1-3.2 GHz. The data recorded by eight antennas around the head is used to form equivalent data from an extended virtual array of 12, 16, and 32 elements. Using quantitative metrics, it is shown that the constructed images from the extended virtual array are more accurate than the images created only from the real antennas. It is also shown that a virtual array that has twice the number of elements of the real array, which meet the minimum limit of degree-of-freedom of the problem, is enough to generate an accurate image with optimized computational resources. In comparison with existing correlation-based methods, the presented approach provides more accurate images.