Quantitative Susceptibility Mapping through Model-based Deep Image Prior (MoDIP)

TL;DR

MoDIP introduces a training-free, model-based deep image prior method for QSM dipole inversion, demonstrating superior generalization, robustness, and efficiency over supervised and traditional methods.

Contribution

It proposes a novel unsupervised, training-free approach combining a small untrained network with physical model enforcement for QSM reconstruction.

Findings

Achieves over 32% accuracy improvement over supervised methods.

33% more computationally efficient than traditional approaches.

Runs 4 times faster, enabling quick 3D high-resolution imaging.

Abstract

The data-driven approach of supervised learning methods has limited applicability in solving dipole inversion in Quantitative Susceptibility Mapping (QSM) with varying scan parameters across different objects. To address this generalization issue in supervised QSM methods, we propose a novel training-free model-based unsupervised method called MoDIP (Model-based Deep Image Prior). MoDIP comprises a small, untrained network and a Data Fidelity Optimization (DFO) module. The network converges to an interim state, acting as an implicit prior for image regularization, while the optimization process enforces the physical model of QSM dipole inversion. Experimental results demonstrate MoDIP's excellent generalizability in solving QSM dipole inversion across different scan parameters. It exhibits robustness against pathological brain QSM, achieving over 32% accuracy improvement than supervised deep learning and traditional iterative methods. It is also 33% more computationally efficient and runs 4 times faster than conventional DIP-based approaches, enabling 3D high-resolution image reconstruction in under 4.5 minutes.