SU-E-J-109: Accurate Contour Transfer Between Different Image Modalities Using a Hybrid Deformable Image Registration...

Area of Science:

• Medical imaging
• Radiation oncology
• Image processing

Background:

• Accurate contour transfer between medical image modalities is crucial for radiation treatment planning.
• Challenges include low image contrast and significant image deformation, impacting contour accuracy.
• Existing methods often fall short under these difficult conditions.

Purpose of the Study:

• To develop and evaluate a novel tool for improving contour transfer accuracy between different image modalities.
• The tool aims to address challenges of low image contrast and large image deformation.
• Comparison against commonly used contour transfer methods in radiation treatment planning.

Main Methods:

• The tool integrates image intensity adjustment and fuzzy connectedness (FC) segmentation with deformable image registration.
• It processes multi-modality images (CT, MRI) by adjusting intensity distributions and utilizing FC segmentation for contour refinement.
• Deformable registration algorithms (B-spline, demons) and a gradient distance weighting algorithm are employed.

Main Results:

• Automatic contour transfer using multi-modality deformable registration improved accuracy by up to 10% compared to rigid transfer.
• The proposed method, incorporating intensity adjustment and affinity map modification, achieved an average accuracy improvement of 14%.
• Fuzzy connectedness segmentation requires careful parameter initialization and user input for optimal performance.

Conclusions:

• Deformable image registration, enhanced by contrast adjustment and fuzzy connectedness segmentation, significantly improves contour transfer accuracy.
• This approach is particularly effective for multi-modality image analysis with substantial deformation and low contrast.
• The developed tool offers a more accurate solution for contour transfer in radiation treatment planning.