Evaluation of motion correction methods in human brain PET imaging—A simulation study based on human motion data

Abstract

Purpose: Motion correction in PET has become more important as system resolution has improved. The purpose of this study was to evaluate the accuracy of event-by-event and frame-based MC methods in human brain PET imaging. Methods: Motion compensated image reconstructions were performed with static and dynamic simulated high resolution research tomograph data with frame-based image reconstructions, using a range of measured human head motion data. Image intensities in high-contrast regions of interest (ROI) and parameter estimates in tracer kinetic models were assessed to evaluate the accuracy of the motion correction methods. Results: Given accurate motion data, event-by-event motion correction can reliably correct for head motions. The average ROI intensities and the kinetic parameter estimates V T and BP ND were comparable to the true values. The frame-based motion correction methods with correctly aligned attenuation map using the average of externally acquired motion data or motion data derived from image registration give comparable quantitative accuracy. For large intraframe (>5 mm) motion, the framebased methods produced ∼9% bias in ROI intensities, ∼5% in V T , and ∼10% in BP ND estimates. In addition, in real studies that lack a ground truth, the normalized weighted residual sum of squared difference is a potential figure-of-merit to evaluate the accuracy of motion correction methods. Conclusions:The authors conclude that frame-based motion correction methods are accurate when the intraframe motion is less than 5 mm and when the attenuation map is accurately aligned. Given accurate motion data, event-by-event motion correction can reliably correct for head motion in human brain PET studies.

Impact of time-of-flight on indirect 3D and direct 4D parametric image reconstruction in the presence of inconsistent dynamic PET data

“…However, such emission/attenuation mismatch could also occur in the absence of any inter/intra-frame motion as patient motion could occur in the time interval between the transmission and emission acquisitions while the patient remains still during the emission scan. A number of strategies have been devised for motion correction (Jin et al 2013). These include online optical motion tracking systems and incorporation of motion information in image reconstruction (Bloomfield et al 2003, Qiao et al 2006, Rahmim et al 2007, Dinelle et al 2011, post-reconstruction frame-by-frame realignment methods using either optical systems or image-based algorithms (Wardak et al 2010, Costes et al 2009, Mourik et al 2009, Ye et al 2014, Jiao et al 2015, as well as more advanced methods leading to simultaneous estimation of motion and kinetic parameters (Jiao et al 2014).…”

Receptor Occupancy of the -Opioid Antagonist LY2456302 Measured with Positron Emission Tomography and the Novel Radiotracer 11C-LY2795050

“…: 388 6 167 MBq, n 5 37) over 1 minute by an automatic pump (Harvard PHD 22/2000; Harvard Apparatus Holliston, MA). Dynamic scan data were reconstructed in 27 frames (6 Â 0.5, 3 Â 1, 2 Â 2, and 16 Â 5 minutes) with corrections for attenuation, normalization, scatter, randoms, and deadtime using the MOLAR algorithm Jin et al, 2013). Motion correction was included in the reconstruction algorithm based on measurements with the Polaris Vicra sensor (NDI Systems, Waterloo, Canada) with reflectors mounted on a swim cap worn by the subject.…”

Direct reconstruction of parametric images for brain PET with event-by-event motion correction: evaluation in two tracers across count levels

“…Motion correction is critical for dynamic brain PET of awake subjects [9, 10]. However, there are very few examples in the literature for which direct reconstruction of brain data has included motion correction.…”

“…The work of Jin et al. [10] suggests that, while frame-based motion correction is sufficient for small intra-frame motions (< 5 mm), event-by-event motion correction has universally good performance.…”

“…Most published demonstrations of direct reconstruction use sinogram data (e.g., [13–15]), but list mode data is better suited for event-by-event motion correction since continuous motion information can be used [10]. When using list mode data, exact computation of the sensitivity image – the sum of the system matrix, including attenuation and normalization, over all possible lines-of-response (LOR) – is difficult since the locations of the lines-of-response are continuously changing with motion.…”