Joint Estimation Attenuation Correction for Respiratory Motion Artifacts

The patent segments the image space into three distinct regions based on lung border information: lung region, border region, and non-lung region. Different reconstruction techniques are applied to each segment - standard attenuation correction for non-lung regions, joint estimation for border regions, and activity estimation for lung regions. This segmentation allows the system to improve accuracy in motion-affected areas without applying complex algorithms uniformly across the entire image, thus balancing accuracy improvement with computational complexity management.

The patent applies different reconstruction qualities to different spatial locations based on their susceptibility to respiratory motion. Border regions surrounding the lungs, which are most affected by motion, receive the most sophisticated joint attenuation-activity estimation treatment. Lung regions receive activity estimation, while stable non-lung regions receive standard attenuation correction. This local quality approach ensures computational resources are concentrated where they provide the most benefit, improving overall accuracy without uniformly increasing complexity.

The patent divides the emission data into three segments based on spatial location relative to lung borders: lung voxels, border voxels, and non-lung voxels. Only border voxels undergo the computationally intensive joint attenuation-activity estimation process, while lung and non-lung voxels use faster estimation methods. This segmentation strategy concentrates computational effort on the most critical motion-affected regions, achieving improved quantitation accuracy where needed while maintaining fast processing for the majority of the image volume.

The patent applies the sophisticated joint estimation technique partially - only to border regions where it provides the most value - rather than excessively applying it to the entire image. This partial action approach achieves the necessary improvement in measurement precision for motion-affected areas without the prohibitive computational cost of universal application, optimizing the trade-off between accuracy and processing time.

The patent segments the image based on lung border detection and applies different processing strategies to each segment. Border regions, which are most susceptible to respiratory motion artifacts, receive joint attenuation-activity estimation that accounts for motion effects. This segmentation approach improves reliability in the most problematic regions without requiring complex motion correction for the entire image, thus improving image quality with moderate increases in processing complexity.

The patent enhances image quality locally in border regions by applying joint estimation techniques specifically where respiratory motion causes mismatches between CT and PET data. Non-lung regions maintain standard attenuation correction since they are less affected by motion. This local quality improvement strategy increases reliability where it matters most while keeping overall processing complexity manageable.