Experimental Evaluation of a Deformable Registration Algorithm for Motion Correction in PET-CT Guided Biopsy

Area of Science:

• Medical Imaging
• Radiology
• Image Processing

Background:

• Positron emission tomography computed tomography (PET-CT) is vital for guiding percutaneous biopsies.
• Respiratory and cardiac motion artifacts in PET-CT compromise tumor localization, shape accuracy, and attenuation correction.
• Motion artifacts necessitate advanced correction techniques for improved diagnostic and therapeutic accuracy.

Purpose of the Study:

• To develop and evaluate a motion correction method for PET-CT imaging.
• To assess the efficacy of deformable registration algorithms for aligning respiratory-gated CT images.
• To compare different optimization strategies for motion correction in PET-CT.

Main Methods:

• Utilized respiratory-gated CT images from 7 patients undergoing PET-CT scans.
• Implemented and compared two deformable registration algorithms: B-spline and symmetric forces Demons.
• Evaluated two optimization approaches: single reference time point registration and adjacent time point registration with composition.

Main Results:

• The B-spline registration algorithm demonstrated superior performance in correcting motion artifacts.
• The reference optimization approach, when combined with B-spline registration, yielded the best overall results.
• Accurate alignment of CT images across the respiratory cycle was achieved, reducing motion-related distortions.

Conclusions:

• Deformable registration of respiratory-gated CT images is an effective strategy for PET-CT motion correction.
• The B-spline algorithm with reference optimization offers a robust solution for enhancing PET-CT image accuracy.
• This method has the potential to improve tumor localization and treatment planning in oncological imaging.