Impact of partial-volume correction in oncological PET studies: a systematic review and meta-analysis
Matthijs C F Cysouw1, Gerbrand M Kramer1, Linda J Schoonmade2
1Department of Radiology and Nuclear Medicine, VU University Medical Centre, P.O. Box 7057, 1007 MB, Amsterdam, Netherlands.
Purpose:
Positron-emission tomography can be useful in oncology for diagnosis, (re)staging, determining prognosis, and response assessment. However, partial-volume effects hamper accurate quantification of lesions <2-3× the PET system's spatial resolution, and the clinical impact of this is not evident. This systematic review provides an up-to-date overview of studies investigating the impact of partial-volume correction (PVC) in oncological PET studies.
Methods:
We searched in PubMed and Embase databases according to the PRISMA statement, including studies from inception till May 9, 2016. Two reviewers independently screened all abstracts and eligible full-text articles and performed quality assessment according to QUADAS-2 and QUIPS criteria. For a set of similar diagnostic studies, we statistically pooled the results using bivariate meta-regression.
Results:
Thirty-one studies were eligible for inclusion. Overall, study quality was good. For diagnosis and nodal staging, PVC yielded a strong trend of increased sensitivity at expense of specificity. Meta-analysis of six studies investigating diagnosis of pulmonary nodules (679 lesions) showed no significant change in diagnostic accuracy after PVC (p = 0.222). Prognostication was not improved for non-small cell lung cancer and esophageal cancer, whereas it did improve for head and neck cancer. Response assessment was not improved by PVC for (locally advanced) breast cancer or rectal cancer, and it worsened in metastatic colorectal cancer.
Conclusions:
The accumulated evidence to date does not support routine application of PVC in standard clinical PET practice. Consensus on the preferred PVC methodology in oncological PET should be reached. Partial-volume-corrected data should be used as adjuncts to, but not yet replacement for, uncorrected data.
