Premium
WE‐D‐L100J‐03: Model Based Respiratory Motion Compensation Using Static PET Acquisition
Author(s) -
Qiao F,
Pan T,
Clark J,
Mawlawi O
Publication year - 2007
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2761540
Subject(s) - scanner , imaging phantom , motion compensation , computer vision , physics , noise (video) , image quality , medical imaging , iterative reconstruction , artificial intelligence , artifact (error) , computer science , nuclear medicine , optics , image (mathematics) , medicine
Purpose: Model based respiratory motion compensation in PET imaging usually assumes PET data to be acquired in gated mode and then models motion during the image reconstruction process (gated‐MC). In this abstract we introduce and evaluate a new approach whereby motion compensation is performed on static PET data while an overall motion is incorporated into the system model (static‐MC). Method and Materials: A phantom consisting of two hot spheres placed in a warm background was scanned using the GE DST PET/CT scanner. One sphere was stationary while the other moved sinusoidally (2cm excursion, 5sec period) in the axial direction during the scan. A 4D‐CT was then performed followed by a list‐mode PET scan of 3 minutes, with gating signals injected into the list‐data stream. The PET data were then unlisted into a single static as well as 10 gated sinograms. Motion was either estimated by registering the 4D‐CT image sequence (RM) or derived from the true sinusoidal motion (TM). Images were reconstructed using OSEM and the following combinations of motion and acquisition schemes were compared: 1) TM+gated; 2) RM+gated; 3) TM+static; 4) RM+static. Evaluation was done by visual inspection and line profiles. In addition, a plot of contrast recovery rate (CRR) versus noise was generated by changing the number of reconstruction iterations. Results: Both static‐MC and gated‐MC approaches showed similar CRR at the same noise level. Visual inspection and line profiles showed that when using TM, the Gated‐MC and Static‐MC had similar image qualities; however with RM, the Gated‐MC performed better. Furthermore, the change in CRR when using RM versus TM was larger for the static‐MC when compared to the gated‐MC approach (3.1% static‐MC compared to 0.2% gated‐MC). Conclusion: Motion compensation without gating PET acquisition is feasible. However, the static‐MC approach is more sensitive to inaccuracies in motion estimation.