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

Summary

Xu, Bagci, Gao and Mollura describe an iterative method for partial volume correction (PVC)—an approach to addressing measurement errors caused by limited spatial resolution—in positron emission tomography (PET). The method estimates uptake regions using affinity-propagation segmentation, applies segmentation-informed regional-mean denoising, and performs voxel-wise correction. It weights successive corrections according to shape consistency between segmentations, iterating until the segmentation converges. The authors evaluated the method on 10 PET/CT images of a NEMA phantom and compared it with deconvolution.

• The authors report qualitative and quantitative phantom results. The evaluation does not establish clinical readiness or prospective performance.
• The paper notes that its conservative strategy for small, low-contrast lesions may avoid over-correction and false positives, but says that correctly identifying such lesions and correcting them without overestimation remains future work.

Publication details

Ziyue Xu, Ulas Bagci, Mingchen Gao, Daniel J. Mollura. “Highly precise partial volume correction for PET images: An iterative approach via shape consistency.” In 2015 IEEE 12th International Symposium on Biomedical Imaging (ISBI 2015), pp. 1196–1199. IEEE, 2015. DOI: 10.1109/ISBI.2015.7164087.