Improved attenuation correction registration for FDG PET/CT images using data-driven gating (DDG)-based motion match
Zoë Wilson1, Manjeet Kuhar1, Kuan-Hao Su2
1Department of Medical Physics and Clinical Engineering, Oxford University Hospitals Foundation Trust, Oxford, UK.
Background:
Advancements in PET technologies and reconstruction methods have improved spatial resolution and noise in PET/CT, making respiratory motion artefacts more impactful on image quality. The motion-match CT (MMCT) algorithm estimates the phase of a helical CT and warps it towards the end expiration of a patient's respiratory cycle. This MMCT-attenuation correction (MMCTAC) can then be used to reconstruct whole body FDG PET/CT data. The performance of these images were compared to those reconstructed using the uncorrected clinical standard attenuation correction from helical CT (normAC), indicating the proportion of cases where MMCTAC images could reduce motion artefact and improve lesion detectability.
Methods:
A sequential cohort of 145 whole-body FDG PET/CT scans was previously evaluated for a data-driven gating study, which identified the gated region of interest (ROI) in 98 high-motion patients. 23 of these high-motion patients had small lesions (n=36) identified in the ROI by a clinician to undergo qualitative analysis comparing the two attenuation correction maps used. PET data were reconstructed using BSREM iterative reconstruction with 3 different portions of the collected data: 1. Ungated 6 min, 2. Ungated 3 min, 3. Quiescent period gated. These reconstructions were generated using normAC or MMCTAC, totaling six reconstructions for semi-quantitative analysis and clinician evaluation. An experienced radiologist ranked the 6 images and scored them on a 5-point Likert Scale for lesion detectability, diagnostic confidence, and image quality.
Results:
In 52% (n=12) of patients with lesions, the clinician ranked all three MMCTAC images higher than the three normAC equivalent reconstructions. Lesion detectability and diagnostic confidence scores retained positive mean differences across the three MMCTAC reconstructions compared to normAC. Similarly, the SUVmax values for 36 lesions demonstrated a significant increase (p < 0.05) for MMCTAC versus normAC images across all three reconstructions.
Conclusions:
The MMCT algorithm successfully enables the phase-matching of a helical CT to the PET data to reduce the presence of respiratory motion artefact for FDG PET/CT images. Overall, the motion-matched CTAC image scores demonstrated the algorithm's efficacy at reducing the motion artefacts present in FDG PET/CT images, improving diagnostic capability for assessment of lesions in the thorax and upper abdomen.
