Integrated Multi-Task MR for Abdominal Radiation Therapy: Towards Adaptive Planning and Treatment Monitoring
J Chen1,2*, P Han2,3, Y Tian1, W Yang1, Y Yang2, D Li3, A Christodoulou3, K Nayak1, Z Fan1, (1) University of Southern California, Los Angeles, CA, (2) UCLA, Los Angeles, CA, (3) Cedars-sinai Medical Center, Los Angeles, CA,
Presentations
TH-B-206-3 (Thursday, 7/14/2022) 8:30 AM - 9:30 AM [Eastern Time (GMT-4)]
Room 206
Purpose: MR-guided radiation therapy (MRgRT) holds great promise for precise radiation dose delivery by presenting real-time images with superior soft tissue contrast. In abdominal RT, however, the treatment plan developed via a prior imaging simulation session often needs quick adaptation on the treatment day due to the position and shape variations. On an MR-Linac, this will mandate multiple MR scans with variable acquisition protocols for pre-beam simulations and real-time treatment monitoring, introducing complexity and inter-scan inconsistency. In this study, we propose an integrated multi-task MR framework to 1) generate volumetric, multi-contrast and motion-resolved images for adaptive radiation planning and 2) enable real-time, volumetric, and simultaneous multi-contrast display for treatment monitoring – all from a single MR protocol.
Methods: The multi-task MR sequence consists of repetitive cycles of saturation pulses, three-stage FLASH readouts, and T2 preparation pulses to generate T1, T2 and PD weighted contrast. A Cartesian spiral-in sampling pattern is used with a navigator (Ky=Kz=0) line acquired every 10 readouts. From a 9-min pre-beam scan, using MR Multitasking, we calculate a spatial basis and a temporal tensor to generate the multi-dimensional images for adaptive planning. Additionally, a linear relationship between the navigator and the temporal tensor can be established. During the treatment scan, real-time multidimensional images can be generated on-the-fly with multiplication operation of the pre-learned spatial basis and real-time navigator lines. We developed this framework on a 3T scanner and tested its feasibility at 0.55T.
Results: At 3T, we demonstrated our capabilities in producing motion-resolved, volumetric multi-contrast images and enabling real-time image generation and display. At 0.55T, we also demonstrated the feasibility of the integrated framework, with temporal resolution as 90ms and latency as 40ms.
Conclusion: Multi-task MR is an integrated framework for adaptive abdominal MRgRT. Investigations of its clinical utility is underway.
Keywords
Taxonomy
IM/TH- MRI in Radiation Therapy: Development (new technology and techniques)
Contact Email