MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Development and Evaluation of General Simultaneous Motion Estimation and Image Reconstruction (G-SMEIR)

S Zhou1*, Y Chi1, J Wang2, M Jin1, (1) The University of Texas at Arlington, Arlington, TX, (2) UT Southwestern Medical Center, Dallas, TX

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: To reconstruct better 4D con-beam CT (CBCT) images, we further develop and thoroughly evaluate general simultaneous motion estimation and image reconstruction (G-SMEIR) that combines the projection domain and the image domain motion estimation.


Methods: Recently, we proposed G-SMEIR for 4D CBCT to overcome the local optimum trapping problem of SMEIR. In G-SMEIR, SMEIR is done for every phase as the reference phase and motion estimation in the image domain is used to update the deformation vector fields to initiate next round of SMEIR. Using different numbers of image domain and projection domain motion estimation, G-SMEIR collapses down to either SMEIR (with projection domain motion estimation only) or 4D reconstruction with the image domain motion estimation only (IM4D). Based on the quantitative metrics, such as root mean squared error (RMSE) and structural similarity index (SSIM), we first identified the best combination of projection domain and image domain motion estimation with fixed projection/backprojection operations. Then, we thoroughly evaluate G-SMEIR quantitatively using 4D XCAT phantom and qualitatively using a real patient CBCT data.


Results: Two times of image domain motion estimation and 22 times of projection domain motion estimation are best for G-SMEIR using fixed 24 projection/backprojection operations. For the tumor motion recovery of XCAT, both SMEIR and G-SMEIR outperform IM4D for large motion in superior-inferior (S-I) direction. Furthermore, G-SMEIR improves over SMEIR in anterior-posterior (A-P) and left-right (L-R) motions. The average RMSE and SSIM over 10 phases are 9.02x10?4 and 0.9648 for G-SMEIR, which are much better than SMEIR using any of phase as the reference phase. The differences are statistically significant using two-sample t-test (p < 0.05). For the patient data, G-SMEIR provide less blurred images than SMEIR, which leads to a better-defined tumor.


Conclusion: G-SMEIR significantly improved SMEIR for 4D CBCT reconstruction based on simulation and patient results.

Funding Support, Disclosures, and Conflict of Interest: This work was supported in part by the U.S. National Institutes of Health under Grant No. NIH/NCI R15CA199020-01A1 and NIH/NIBIB R03EB021600-01A1.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email