Highly precise partial volume correction for PET images: An iterative approach via shape consistency

Positron emission tomography (PET) is capable of capturing the functional information. A major limitation for PET imaging is the low spatial resolution, leading to partial volume effects (PVE). PVE introduces significant bias to the image quantification, causing compromised measurement for uptake regions, especially smaller ones. For quantitative PET, accurate uptake values are critical for diagnostic evaluation and treatment planning. Therefore, a partial volume correction (PVC) technique is highly desirable in order to avoid size-dependent underestimation for true activities. In this paper, we present a new iterative PVC approach for PET images. The proposed method uses the state-of-the-art simultaneous delineation and noise removal algorithm to estimate the local uptake regions. The delineation is further utilized for weighted PVC with regard to a shape consistency measurement. The process is performed iteratively until delineation convergence. Qualitative and quantitative results demonstrate that the proposed framework successfully corrects the PVE and preserves local structures.