Click here to


Are you sure ?

Yes, do it No, cancel

Challenges of Attenuation Correction in Conventional Respiratory Motion Corrected PET/CT

O Mawlawi1 , J Meier2*, (1) MD Anderson Cancer Ctr., Houston, TX, (2) UT MD Anderson Cancer Center, Houston, TX


(Tuesday, 7/31/2018) 3:45 PM - 4:15 PM

Room: Exhibit Hall | Forum 8

Purpose: Respiratory motion in PET/CT is unavoidable, resulting in underestimated activity concentration, and misalignments with the CT used for attenuation correction(CTAC). One recent implementation of commercial motion correction reconstructs gated PET images, elastically registers them, and averages the images(Q-Freeze), and requires a phase matched CT for proper attenuation correction. Another form of motion correction, quiescent phase gating(Q-Static) which uses only data from the quiescent phase, and non-motion corrected STATIC PET images are typically acquired with a free breathing 3D_CTAC. In this work we determine which method of attenuation correction and motion correction yield superior image quality and quantification.

Methods: A retrospective analysis was performed on 5 patients with lung or liver lesions of diameter between 1-3 cm acquired on a GE Discovery 690 PET/CT. For the Q-Freeze reconstruction, 6 gates were used, and attenuation correction was performed with a phase matched QF_4DCT. For Q-Static, the PET data was selected from phase range 35-75%, retaining 40% of the data. For STATIC, all data was used with no motion correction. Two attenuation correction CTACs were used for Q-Static and STATIC, one with End expiration(EXP) and End Inspiration(EI). For all methods, a 7.5 minute PET LIST dataset was used. For each patient, motion compensation method, and acquisition time, we measured tumor SUVmax and the SUV standard deviation in the healthy liver.

Results: For the 7 tumors analyzed, the average(+/- stdev) SUVmax listed in decreasing order was, Q-Static_EXP(8.95+/-4.26), Q-Freeze_4DCT(7.9+/-3.55), Q-Static_EI(6.9+/-2.68), STATIC_EXP(6.57+/-2.79), STATIC_EI(5.42+/-1.95). The average(+/- stdev) of SUV standard deviation listed in decreasing order was, Q-Static_EI(0.2+/-0.11), Q-Static_EXP(0.16+/-0.02), STATIC_EI(0.16+/-0.16), Q-Freeze_4DCT(0.16+/-0.04), STATIC_EXP(0.12+/-0.06).

Conclusion: Q-Static_EXP results in the highest SUVmax due to being in best alignment with the AC map, and also due to having higher image noise. Q-Freeze and STATIC use all of the acquired PET data, resulting in lower image noise and consequentially, lower SUVmax.

Funding Support, Disclosures, and Conflict of Interest: Research Agreement with Siemens Healthineers


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email