Ultrafast Volumetric Cine Magnetic Resonance Imaging (UVC-MRI) for Real-Time 3-Dimensional Target Localization/Tracking

Abstract

Recently we developed a VC MRI technique to generate quasi-real-time (3 frames/s) 3D MRI based on motion modeling and 2D cine MRI for target localization. In this study, we hypothesize that the frame rate of VC MRI can be substantially accelerated using sparsely sampled 2D cine MRI. This Ultrafast VC MRI (UVC MRI) can minimize the motion artifacts for irregular breathing and provide truly real-time 3D guidance for gating/target tracking. The purpose of this study is to validate the feasibility of UVC MRI and evaluate the effects of different scanning parameters on UVC MRI. The 4D MRI acquired during patient simulation are used as prior images. Principal component analysis is used to extract 3 major respiratory deformation patterns from the deformation field maps (DFM) registered between end expiration phase and all other phases of the 4D MRI. Onboard UVC MRI at any instant is considered as a deformation of the prior MRI at the end expiration phase. The DFM for UVC MRI is represented by a linear combination of the 3 major deformation patterns. Coefficients of the deformation patterns are solved by matching the corresponding 2D slice of UVC MRI with the acquired onboard 2D cine MRI. The ultrafast onboard 2D cine MRI are acquired at 30 frames/s by sampling only 10% of the k-space on Cartesian grid, with 85% of that taken at the central k-space and the other 15% randomly sampled elsewhere. UVC MRI was evaluated using both XCAT (computerized patient model) simulation of lung cancer patients and MRI data from liver cancer patients. Three scenarios were simulated from prior to onboard volume in XCAT, including phase shift, synchronous and asynchronous motion amplitude change between tumor and body motion. The accuracy of UVC MRI was evaluated using volume percent difference (VPD) and center of mass shift (COMS) of the estimated tumor volume. Effects of region of interest (ROI) selection, 2D cine slice orientation, slice location, and slice number on the estimation accuracy were evaluated. In XCAT study, the UVC MRI estimated using a single sparsely sampled sagittal 2D cine MRI with a ROI around tumor achieved VPD/COMS of 8.54±2.27%/0.24±0.05 mm among all scenarios. Using the entire cine image instead of ROI reduced the accuracy to 37.37±43.45%/4.03±6.02 mm. Changing the 2D cine orientation to axial or coronal view reduced the accuracy to 9.26±1.30%/0.76±0.72 mm and 15.56±2.91%/1.80±0.56 mm, respectively. Estimation using the sagittal cine was the most robust against slice location change among all views. Using multiple parallel or orthogonal cine images did not further improve UVC MRI estimation as single cine was sufficient to fit the motion model. In patient study, profiles in the estimated and ground truth UVC MRI agreed within 5% error. Preliminary studies showed it is feasible to generate UVC MRI up to 30 frames/s to provide truly real-time 3D target verification. UVC MRI can potentially minimize/eliminate treatment errors in liver or lung SBRT to improve treatment outcome.