A subject-specific unsupervised deep learning method for quantitative susceptibility mapping using implicit neural representation

Ming Zhang, Ruimin Feng, Zhenghao Li, Jie Feng, Qing Wu, Zhiyong Zhang, Chengxin Ma, Jinsong Wu, Fuhua Yan, Chunlei Liu, Yuyao Zhang, Hongjiang Wei

Abstract: Highlights•Subject-specific and unsupervised deep learning for QSM reconstruction.•Integration of implicit continuous signal representation and explicit regularizations.•Phase compensation strategy for an accurate physical model.•Improved accuracy and quality compared with established methods.