Data-driven event-by-event respiratory motion correction using TOF PET list-mode centroid of distribution

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Positron Emission Tomography (PET)

Background:

• Respiratory motion significantly degrades PET image quality.
• External motion tracking systems are commonly used for correction but can be cumbersome.
• Data-driven methods offer a promising alternative for motion correction without external devices.

Purpose of the Study:

• To develop and evaluate a data-driven, event-by-event respiratory motion correction technique using the Centroid-of-distribution (COD) algorithm for Time-of-Flight (TOF) PET.
• To compare the performance of the COD-based method with an external system-based (Anzai belt) correction method.
• To assess the impact of data-driven motion correction on PET image quality.

Main Methods:

• Utilized the Centroid-of-distribution (COD) algorithm on TOF PET list-mode data to derive respiratory motion traces.
• Compared COD-derived motion traces with those from an Anzai belt system using Pearson correlation coefficients.
• Performed gated reconstructions using both COD and Anzai traces to evaluate motion capture accuracy.
• Implemented event-by-event motion correction within the MOLAR reconstruction framework using COD traces.

Main Results:

• Data-driven COD traces showed good correlation (63-89%) with Anzai traces in superior-inferior and anterior-posterior directions.
• No significant difference in pancreatic motion displacement was observed between COD-based and Anzai-based gated reconstructions.
• COD-based event-by-event correction achieved comparable contrast recovery and reduced motion blur to the Anzai-based method.
• Significant image quality improvement was noted with COD-based correction compared to uncorrected reconstructions.

Conclusions:

• The data-driven Centroid-of-distribution (COD) algorithm effectively corrects for respiratory motion in TOF PET studies.
• COD-based event-by-event motion correction provides comparable results to external system-based methods.
• This technique enhances PET image quality without the need for external motion tracking systems.