Abstract
BACKGROUND: Low-field MRI provides superior soft-tissue contrast compared to CT while costing significantly less than high-field MRI, which makes it a more accessible option for MRI-guided radiation therapy planning. Four-dimensional MRI (4D-MRI) is a technique that has been increasingly adopted clinically for internal target volume (ITV) delineation in free-breathing liver radiotherapy planning, and it requires high spatial resolution and accurate respiratory phase differentiation to enable precise dose planning. The feasibility of 4D-MRI at low-field strength, specifically at 0.55T, has not been evaluated. PURPOSE: This study aims to investigate the feasibility of 4D-MRI for ITV delineation in liver radiation therapy planning using a commercial 0.55T MRI scanner. METHODS: A 3D stack-of-stars T1-weighted sequence was implemented with two respiratory self-navigation methods: (1) k-space center point tracking ("k-center") and (2) superior-inferior one-dimensional projection-based tracking ("SI-projection"). These methods were evaluated using ten phantom scans simulating diverse respiratory motion patterns and five liver tumor patient scans. RESULTS: Both self-navigation approaches demonstrated strong correlation between the extracted self-gating signals (SGS) and ground-truth motion traces across four breathing patterns: sinusoidal waveform, typical respiration, drifting motion, and irregular breathing. For sinusoidal motion, measured ITV deviations were within 1.1% of the true ITV for both methods. In non-sinusoidal cases, ITV deviations remained within 2% except for two drifting motion cases where k-center SGS based reconstructions showed deviations of 6.0% and 2.4%. In liver tumor patient scans, both self-navigation techniques produced images with sufficient tumor delineation for treatment planning, with SI-projection SGS-based reconstructions yielding sharper images than k-center SGS-based reconstructions. ITV volumes contoured by two radiation oncologists showed strong and comparable inter-observer agreement across both techniques. CONCLUSIONS: This study demonstrates that 4D-MRI at 0.55T is feasible and provides adequate image quality for ITV delineation. Self-navigation techniques play an important role in improving the sharpness of tumor boundaries, with SI-projection based self-navigation offering superior performance in cases of irregular respiratory motion.