Automatic arm removal in PET and CT images for deformable registration

Area of Science:

• Medical Imaging
• Radiology
• Computational Biology

Background:

• Positron Emission Tomography (PET) offers high sensitivity for detecting cancer-related functional abnormalities but lacks detailed anatomical information for precise localization.
• Computed Tomography (CT) provides excellent anatomical detail but is less sensitive to early pathological changes compared to PET.
• Combining PET and CT images through registration enhances cancer localization accuracy and integrates functional and anatomical data.

Purpose of the Study:

• To develop a deformable registration software system for fusing PET and CT images.
• To address the challenge of anatomical differences, specifically the presence and variable positioning of arms in PET versus CT images.
• To introduce an automatic arm removal method to streamline the image fusion process for clinical application.

Main Methods:

• Development of a deformable registration software system utilizing mutual information and a B-spline model.
• Implementation of a novel, automatic algorithm for preprocessing PET/CT images by removing arms.
• Validation of the automatic arm removal method on clinical PET and CT datasets.

Main Results:

• The developed automatic arm removal method effectively handles feature mismatches caused by arm presence and positioning.
• The algorithm demonstrates efficiency and robustness on both clinical PET and CT images.
• Streamlining the registration process through automatic arm removal facilitates smoother image fusion.

Conclusions:

• Automatic arm removal is a crucial preprocessing step for robust deformable registration of PET/CT images.
• This technology enhances the fusion of PET and CT imaging, improving cancer diagnosis, staging, and treatment monitoring.
• The streamlined image fusion process is expected to accelerate the clinical adoption of advanced PET/CT imaging techniques.