Progress In Electromagnetics Research M, Vol. 102, 53–63, 2021 Non-Iterative Microwave Imaging Solutions for Inverse Problems Using Deep Learning Thathamkulam A. Anjit 1, * , Ria Benny 1 , Philip Cherian 2 , and Palayyan Mythili 1 Abstract—This paper describes a U-net based Deep Learning (DL) approach in combination with Subspace-Based Variational Born Iterative Method (SVBIM) to provide a solution for the quantitative reconstruction of scatterer from the measured scattered field. The proposed technique can be used as an alternative to conventional time consuming and computationally complex iterative methods. This technique comprises a numerical solver (SVBIM) for generating the initial contrast function and a DL network to reconstruct the scatterer profile from the initial contrast function. Further, the proposed technique is validated against theoretical and experimental results available from the literature. Root Mean Square Error (RMSE) value is used as the metric to measure the accuracy of the reconstructed image. The RMSE values of the proposed method show a significant reduction in the reconstruction error compared with the recent Back Propagation-Direct Sampling Method (BP-DSM). The proposed method produces an RMSE value of 0.0813 against 0.1070 in the case of simulation (Austria Profile). The error value obtained by validating against the FoamDielExt experimental database in the case of the proposed method is 0.1037 against 0.1631 reported for BP-DSM method. 1. INTRODUCTION Microwave Imaging (MWI) is a promising non-contact and non-destructive imaging technique that has been effectively used for a wide range of engineering and medical applications. Major applications which utilize the MWI are surveillance, microwave remote sensing, non-destruction detection, biomedical imaging, and geological exploration [1, 2]. MWI can be broadly classified into two categories, namely quantitative and qualitative methods. Qualitative methods can only retrieve the shape and location of the scatterers within the imaging domain, hence these techniques are used for detection purposes. On the other hand, the quantitative imaging technique is an electromagnetic inverse scattering problem that aims to reconstruct the spatial permittivities of the scatterer from the knowledge of the measured scattered fields with a few receivers [3, 4]. These Inverse Scattering Problem (ISP) techniques are generally slower, ill-posed, and nonlinear. Imaging such a scatterer from the ISP is achieved by solving the Electric Field Integral Equation (EFIE). These EFIE equations are nonlinear with two sets of unknown quantities (a) the dielectric profile and (b) the total electric field inside the scattering object [5]. This non-linearity occurs due to three major limitations (a) presence of evanescent waves — fine-grained details will be lost in the inversion process since the evanescent waves produced in the region under study do not reach the receiver location, (b) local minima — by increasing the number of fields of excitation and the number of sensors, and local minima(s) can be reduced to some extent [6], and (c) sensitivity of the inverse problem to noise — the effect of noise can be reduced by employing a conventional regularization technique along with the inversion algorithm. Over the past decades, ISP is often formulated as a linear inverse problem by adopting scattering models based on Born iteration method (BIM) and Rytov method which produce iterative solutions to Received 13 February 2021, Accepted 2 April 2021, Scheduled 13 April 2021 * Corresponding author: Thathamkulam Agamanan Anjit (taanjit@yahoo.in). 1 Cochin University of Science and Technology, Cochin, Kerala, India. 2 College of Engineering, Chengannur, Kerala, India.