Program Information
Attenuation and Motion Correction of 4D-PET Images Using Simultaneous Motion Estimation and Image Reconstruction (SMEIR) of 4D-CT Data
F Kalantari*, Y Zhong , J Wang , UT Southwestern Medical Center, Dallas, TX
Presentations
WE-F-201-3 (Wednesday, August 2, 2017) 1:45 PM - 3:45 PM Room: 201
Purpose: Respiratory gated PET followed by post reconstruction registration (PRR) is a potential solution to respiratory-induced motion artifacts. CT based attenuation correction (AC) of individual PET frames is an essential step toward 4D-PET quantification. In this study, we investigated the use of Simultaneous Motion Estimation and Image Reconstruction (SMEIR) method to reconstruct 4D-CT and estimate motion vectors for attenuation and motion correction of 4D-PET.
Methods: The SMEIR algorithm provides two distinct advantages over conventional 4D-CT reconstruction methods. First, it reconstructs any individual CT frame using all projections with reduced streaking artifact and noise, hence improving the PET-AC process. Second, it estimates deformation vector fields (DVFs) by matching the forward projection of the deformed reference CT image and projections of other phases. Accurate DVFs improve PET frames alignment and PRR accuracy. For assessment, XCAT phantoms at 8 different respiratory phases with lung tumors were generated. 20 CT projections were acquired at each phase resulting in a total of 160 projections. The performance of the SMEIR algorithm was quantitatively evaluated by measuring the mean SUV_mean of lung tumors in PET frames with AC using SMEIR-derived CT images as well as PRR volumes using SMEIR-derived DVFs. SUV values were also compared to those from the AC-PET frames using conventional CT reconstructions (FDK and ART-TV) and Demons motion estimation for PRR.
Results: CT artifacts were reduced in SMEIR method compared to FDK and algebraic reconstruction technique (ART). The error in SUV_mean of 30 mm tumor in AC-PET frames were -8.9%±1.5%, -3.5%±1.9% and -0.8%±2.2% for FDK, ART and SMEIR respectively. The corresponding values for 40 mm tumor were -10.0%±1.4%, -2.2%±1.7% and -0.8%±1.4%. For PRR volumes, SMEIR-derived DVFs eliminated artifacts that were visible when Demons DVFs were used.
Conclusion: SMEIR method and 4D-CT projections can provide valuable attenuation and motion information for quantitative 4D-PET imaging.
Contact Email: