Markerless head motion tracking and event-by-event correction in brain PET

Area of Science:

• Medical Imaging
• Neuroscience
• Biomedical Engineering

Background:

• Head motion correction (MC) is crucial for accurate brain PET imaging.
• Current hardware-based motion tracking (HMT) requires head-mounted markers, limiting clinical translation.
• Markerless HMT (MLMT) offers a more convenient alternative for clinical applications.

Purpose of the Study:

• To validate the United Imaging Healthcare motion tracking (UMT) MLMT system for PET head MC.
• To compare UMT performance against the established Polaris Vicra HMT system.
• To introduce and evaluate a novel metric, registration quality (RQ), for assessing motion correction.

Main Methods:

• UMT uses a stereovision camera with infrared structured light for real-time 3D facial surface capture.
• Point clouds are acquired at 30 Hz and registered using a rigid-body iterative closest point algorithm.
• Validation involved phantom and human point source studies, alongside human PET studies with 18F-FPEB and 11C-LSN3172176.

Main Results:

• UMT demonstrated superior reconstruction in phantom studies (0.35 ± 0.27 mm spatial resolution, 0.12 ± 0.10 mm residual displacement).
• Human studies showed UMT achieved comparable spatial resolution to Vicra with lower noise.
• UMT yielded comparable region of interest values and negligible SUV differences in human PET studies.

Conclusions:

• The UMT MLMT system provides effective and accurate head motion correction for brain PET imaging.
• UMT performance is comparable to the established Vicra HMT system, offering a more user-friendly solution.
• The proposed RQ metric effectively evaluates motion correction quality, correlating with existing methods.