Local respiratory motion correction for PET/CT imaging: Application to lung cancer

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Processing

Background:

• Whole-field-of-view respiratory motion correction in PET imaging has limited clinical adoption.
• Local respiratory motion correction (LRMC) offers potential advantages like speed and organ-specific correction.
• Evaluating LRMC with various motion models is crucial for oncology applications.

Purpose of the Study:

• To assess the performance of LRMC for oncology applications, specifically lung lesions.
• To compare different motion models (center of gravity, affine, elastic) within the LRMC framework.
• To evaluate the impact of LRMC on tumor localization and volume accuracy.

Main Methods:

• Utilized simulated (4D cardiac-torso phantom) and clinical (six patients) PET data.
• Defined a Volume of Interest (VOI) on motion-averaged images.
• Reconstructed gated PET images of the VOI and applied center of gravity, affine, or elastic registration to derive motion transformation maps for reconstruction.

Main Results:

• The elastic model, applied locally or whole-FOV, demonstrated superior performance compared to center of gravity and affine models.
• Elastic LRMC significantly improved spatial tumor localization (89% local, 81% whole-FOV) and tumor volume accuracy (84% local, 80% whole-FOV).
• LRMC with a nonrigid deformation model achieved over an order of magnitude gain in computational efficiency compared to whole-FOV correction.

Conclusions:

• LRMC is a flexible and efficient method for correcting respiratory motion effects in single thoracic lesions.
• The elastic registration model within LRMC provides enhanced accuracy for lung lesion PET imaging.
• LRMC shows promise for improving the clinical utility of PET imaging in thoracic oncology.