CycleQSM: Unsupervised QSM Deep Learning using Physics-Informed CycleGAN
This method offers a faster and more accurate QSM reconstruction for medical imaging, particularly beneficial for clinical applications where ground-truth data is scarce and existing methods underestimate susceptibility values.
This paper addresses the ill-posed problem of Quantitative Susceptibility Mapping (QSM) from MRI phase images. The authors propose an unsupervised deep learning method using a physics-informed CycleGAN, which achieves more accurate QSM maps than existing deep learning methods and competitive performance with the best classical approaches.
Quantitative susceptibility mapping (QSM) is a useful magnetic resonance imaging (MRI) technique which provides spatial distribution of magnetic susceptibility values of tissues. QSMs can be obtained by deconvolving the dipole kernel from phase images, but the spectral nulls in the dipole kernel make the inversion ill-posed. In recent times, deep learning approaches have shown a comparable QSM reconstruction performance as the classic approaches, despite the fast reconstruction time. Most of the existing deep learning methods are, however, based on supervised learning, so matched pairs of input phase images and the ground-truth maps are needed. Moreover, it was reported that the supervised learning often leads to underestimated QSM values. To address this, here we propose a novel unsupervised QSM deep learning method using physics-informed cycleGAN, which is derived from optimal transport perspective. In contrast to the conventional cycleGAN, our novel cycleGAN has only one generator and one discriminator thanks to the known dipole kernel. Experimental results confirm that the proposed method provides more accurate QSM maps compared to the existing deep learning approaches, and provide competitive performance to the best classical approaches despite the ultra-fast reconstruction.