Program Information
Employing Compressed Sensing to Improve Tracking Image Quality of the MRI-Guided Radiotherapy System
D Du*, O Pechenaya Green, Y Feng, S Mutic, P Parikh, J Olsen, Y Hu, Washington University School of Medicine, St Louis, MO
SU-C-103-7 Sunday 1:00PM - 1:55PM Room: 103Purpose: To validate the feasibility of using compressed sensing to improve spatial resolution and image quality of motion tracking images for MRI guided radiotherapy system.
Methods: This study was performed on a commercial low field 0.35T MRI-guided radiotherapy system using the ACR phantom. The phantom was moved along the feet-head direction 9 times to mimic motion with relative positions of 0cm, 5mm, 11mm, 15mm, 20mm, 17mm, 13mm, 8mm and 3mm, respectively. For each position, two images were acquired in the sagittal orientation using the standard TrueFISP sequence. One used the standard clinical protocol to acquire images with a spatial resolution of 3.5mm and a temporal resolution of 4 frames per second (FPS). The other was acquired with a higher spatial resolution (2.0mm) but at a lower temporal resolution (0.7 FPS). Compressed sensing (CS) was used to achieve 5.6-fold acceleration and restore the temporal resolution of tracking images to 4 FPS. Images reconstructed using the CS technique were compared qualitatively to the standard technique as well as the low temporal, high spatial resolution images without utilizing the CS technique.
Results: Image quality of the high spatial and high temporal resolution tracking images utilizing the CS technique was very close to the high spatial and low temporal resolution tracking images without utilizing the CS technique, and much better than the tracking images offered by the standard imaging protocol.
Conclusion: Our results confirmed that the CS technique could accelerate the image acquisition speed nearly 6 times while maintaining similar image quality. Therefore, it could be used to significantly improve the spatial resolution of the low-field MRI guided radiotherapy tracking images without sacrificing the temporal resolution.
Contact Email: