Room: AAPM ePoster Library
Purpose: develop and validate a novel Frequency-Modulated (FM) bSSFP MRI approach, which allows images reconstructed at multiple f0 off-resonances, in the application of implanted Gold Fiducial Marker (GFM) visualization in the prostate for radiation therapy planning and tracking. It will improve patient care flow by enhancing GFMs visibility and eliminating time-consuming multiple bSSFP MRI scans with different RF phase-cycling.
Methods: stack-of-stars 3D radial FM-bSSFP sequence was developed and tested on an agar gel phantom with five gold fiducial markers implanted. Sixty-four 3D complex bSSFP datasets were reconstructed at different f0 frequencies using the raw dataset of a single MRI scan. To compensate for the slow background signal change due to global B0 inhomogeneities, a voxel-wise B0 field map was derived from frequency difference at the maximal signal of each voxel. The resulting B0 field map was then low pass filtered and used to generate 64 B0-corrected 3D datasets for FM identification and localization.
Results: complete set of GFMs to background contrast could be generated using multifrequency reconstruction from a single FM-bSSFP scan raw dataset. The on-resonant images had bright background with dark GFMs. The off-resonant reconstructed images had dark background with bright artifacts centered at the GFMs. All GFMs were easily identified on background signal corrected datasets, particularly on images with dark background. Radial k-space sampling also generated symmetric artifacts around the GFMs compared to asymmetric distortion using Cartesian sampling, which further helped improve GFM localization accuracy.
Conclusion: new MRI approach can potentially be used for direct visualization of GFMs in the prostate for MRI-only radiotherapy. Only one single fast FM-bSSFP scan is needed for multiple GFMs to background contrasts through multifrequency reconstruction. The full spectral coverage of each voxel also allows flexible image processing for easy and potentially fully automated GFM localization.