WHOLE BODY NONRIGID CT-PET REGISTRATION USING WEIGHTED DEMONS

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Image Registration

Background:

• Accurate registration of multi-modal images (CT and PET) is crucial for preclinical research.
• Inherent differences in imaging protocols and acquisition times lead to significant nonrigid anatomical changes and heterogeneous image characteristics.
• Existing registration methods struggle with these complex deformations and variations.

Purpose of the Study:

• To present a new deformable registration method for whole-body rat CT and PET images.
• To address challenges posed by nonrigid changes and heterogeneous image characteristics between separate acquisitions.
• To improve the accuracy and robustness of image registration in preclinical settings.

Main Methods:

• A weighted demons algorithm was developed for deformable image registration.
• Utilized both transmission-PET and emission-PET images during the registration process.
• Emphasized specific regions of the transmission-PET image using emission-PET information.
• Validated the method on nine rat image datasets using the M-Hausdorff distance.

Main Results:

• The proposed weighted demons algorithm demonstrated improved registration performance.
• Achieved higher accuracy compared to standard Demons and nonrigid FFD registration methods.
• Quantitatively validated results using the M-Hausdorff distance similarity measure.

Conclusions:

• The weighted demons algorithm offers a robust and accurate solution for registering rat CT and PET images.
• This method effectively handles nonrigid transformations and heterogeneous image properties.
• The approach shows significant potential for enhancing preclinical imaging analysis and research.