MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Dynamic MRI Reconstruction Using Simultaneous K-Space-Driven Motion Estimation and Compensation (SK-MEC)

Y Zhang1*, Z Iqbal1 , C Shen1 , C Wang2 , S Jiang1 , J Wang1 , (1) UT Southwestern Medical Ctr at Dallas, Dallas, TX, (2) Duke University Medical Center, Durham, NC

Presentations

(Tuesday, 7/16/2019) 4:30 PM - 6:00 PM

Room: 225BCD

Purpose: To develop a MRI reconstruction technique (SK-MEC) by combining k-space-driven inter-respiratory-phase motion estimation with motion-compensated image reconstruction.

Methods: The proposed SK-MEC algorithm alternates between two steps: (1) A motion model estimation step to update deformation-vector-fields (DVFs) between a reference phase MRI and other phases; The DVFs are estimated through an inverse-consistent scheme by matching the k-space data of the deformed MR images to the acquired k-space data; and (2). Motion-compensated reconstruction of the reference phase MRI using the updated motion model and the entire k-space data. Upon convergence, the inter-phase DVFs are applied onto the reference phase MRI to generate the full dynamic MRI set. We evaluated SK-MEC using the dynamic extended-cardiac-torso (XCAT) phantom and a dynamic liver patient MRI set. For both, the golden-angle radial sampling scheme was employed to acquire the k-space data with 5, 10 or 20 spokes per respiratory phase, corresponding to an imaging acceleration rate of 80, 40 and 20. The reconstruction accuracy of SK-MEC was compared with the nonuniform-fast-Fourier-transform (NUFFT) technique, the conjugate gradient algorithm with total variation regularization (CG-TV), and the extra�dimensional golden�angle radial sparse parallel (XD-GRASP) technique. The image quality was quantitatively evaluated using the signal-to-error-ratio (SER) metric.

Results: By using 5 spokes per phase, the mean (±s.d.) SERs of reconstructed images by NUFFT, CG-TV, XD-GRASP and SK-MEC were -5.65(±1.30) dB, 9.26(±0.82) dB, 9.79(±0.29) dB and 11.20(±1.02) dB respectively. By using 20 spokes, the corresponding SERs were 3.01(±0.88) dB, 16.04(±1.05) dB, 16.82(±0.65) dB and 17.32(±1.47) dB. The inter-phase DVFs solved by SK-MEC accurately tracked the XCAT liver tumor centroid to a mean error of 2.3 mm by using 5 spokes per phase, and 0.9 mm by using 20 spokes.

Conclusion: SK-MEC allows substantial k-space under-sampling for dynamic MRI acceleration. The solved intra-phase motion can also benefit target localization and adaptive radiotherapy.

Funding Support, Disclosures, and Conflict of Interest: We acknowledge funding support from the American Cancer Society (RSG-13-326-01-CCE), from the US National Institutes of Health (R01 EB020366), and from the Cancer Prevention and Research Institute of Texas (RP130109).

Keywords

Deformation, Reconstruction, MRI

Taxonomy

IM/TH- MRI in Radiation Therapy: General (most aspects)

Contact Email