A method of image registration for small animal, multi-modality imaging

Area of Science:

• Preclinical Imaging
• Biomedical Research
• Medical Physics

Background:

• Multi-modality imaging is essential for in vivo biomedical research.
• Robust registration of images from different tomographic scanners is a routine challenge.
• Existing methods may lack the precision required for high-throughput preclinical studies.

Purpose of the Study:

• To develop and validate a hardware-based registration method for preclinical positron emission tomography (PET) and microcomputed tomography (microCT) imaging.
• To minimize image misalignment errors between different imaging modalities.
• To assess the impact of registration accuracy on quantitative positron emission tomography (PET) data.

Main Methods:

• Implementation of a specialized imaging chamber for rigid mounting on microPET and microCT scanners.
• Design of a 3D grid phantom for software-based spatial transformation matrix generation using a 15-parameter perspective model.
• Synergistic use of the imaging chamber and phantom to achieve accurate image registration.

Main Results:

• Achieved an average registration error of 0.335 mm between microPET and microCT modalities, verified using an in vivo mouse bone scan.
• Estimated potential PET quantitation errors to be less than 5% even with image misalignment.
• Demonstrated the robustness and suitability of the method for routine high-throughput animal imaging.

Conclusions:

• The developed hardware registration method significantly enhances the accuracy of multi-modality preclinical imaging.
• This technique provides a reliable solution for precise image alignment in high-throughput research facilities.
• The method ensures high-quality quantitative PET data, supporting advanced in vivo biomedical investigations.