SU-E-J-99: Automated Registration Method Based on Multi-Scale Edge Preserving Scale Space for PET-CT Scanner.

Area of Science:

• Medical Imaging
• Image Processing
• Computational Anatomy

Background:

• Mutual Information (MI) based registration is crucial for aligning PET-CT images.
• Traditional MI methods struggle with high computational complexity and local optima due to limited spatial information.

Purpose of the Study:

• To enhance Mutual Information (MI) based registration for PET-CT scanners.
• To address challenges of computational complexity and local optima in existing methods.

Main Methods:

• Proposed a novel multi-scale registration framework: Edge Preserving Multi-scale Registration (EPMR).
• Utilized an edge-preserving total variation L1 norm (TV-L1) scale space representation, prioritizing edge size over intensity.
• Developed an optimal TV-L1 parameter estimation for automated registration.

Main Results:

• EPMR demonstrated superior accuracy and robustness compared to traditional methods in multi-modal medical datasets.
• Achieved a 10%-14% increase in registration speed for PET-CT scans.
• Clinical application showed 90% overlap in adaptive Gross Tumor Volume (GTV) re-contouring for radiotherapy.

Conclusions:

• Introduced a novel scale space decomposition methodology for improved PET-CT image registration.
• The EPMR framework enhances the performance of MI-based registration by incorporating meaningful spatial information.