Automated method for small-animal PET image registration with intrinsic validation

Area of Science:

• Medical Imaging
• Neuroscience
• Radiochemistry

Background:

• Accurate image registration is crucial for analyzing small-animal Positron Emission Tomography (PET) studies.
• Existing methods may require significant user intervention and lack objective validation.

Purpose of the Study:

• To propose and compare different registration strategies for aligning small-animal PET data.
• To introduce objective metrics for validating registration accuracy and consistency.

Main Methods:

• An information theory-based registration algorithm was employed.
• Different reference image masking techniques were evaluated.
• Registration consistency was used to detect erroneous alignments.
• The methodology was tested on a dataset of [18F]FDG-PET rat brain images.

Main Results:

• A multiresolution, two-step registration approach demonstrated superior performance.
• Optimal strategy involved using the whole image at low resolution and masking the brain at high resolution.
• This method achieved 87.5% registration success rate and 0.67-mm average accuracy.

Conclusions:

• The proposed registration method offers high accuracy and robustness for small-animal PET studies.
• Automatic error assessment eliminates the need for subjective visual inspection.
• Facilitates efficient and objective analysis of large cohorts in preclinical research.