A 3D Magnetospheric CT Reconstruction Method Based on 3D GAN and Supplementary Limited‐Angle 2D Soft X‐Ray Images

Abstract

AbstractSoft X‐ray imaging is a new method used to observe the structure of the Earth's magnetosphere. Reconstruction of the Earth's three‐dimensional (3‐D) magnetospheric structure from 2‐D soft X‐ray images is an important research topic in astronomical imaging and exploration. In this study, computed tomography (CT) is investigated for application to the planned Solar wind Magnetosphere Ionosphere Link Explorer (SMILE), where resulting images are collected at varying small angles. The restricted geometry of SMILE results in sparse projection data and reduced reconstruction quality. As such, a new algorithm is proposed, based on cone‐beam computed tomography (CBCT) and utilizing 3‐D generative adversarial networks (GAN) to supplement the 2‐D soft X‐ray images and increase the angular coverage of the projections. A complement optimization backpropagation network (COBN) is then used to reconstruct high‐quality 3‐D X‐ray emissivity profiles in the magnetosphere. A series of validation experiments were conducted in which the reconstructions were compared with synthetic results generated by the PPMLR‐MHD (an extended LagRangian version of the Piecewise Parabolic Method –MagnetoHydroDynamic) model under varying solar wind conditions. Results showed the supplemented profiles were more complete and accurate than the direct reconstructions, as measured by the peak signal‐to‐noise ratio (PSNR) and structural similarity (SSIM). This suggests the proposed technique could be a viable approach to observing magnetosphere structure and variability.