Abstract
1743
Objectives: The accuracy of PET image quantitation is severely affected by partial volume effect (PVE) caused by limited spatial resolution of PET tomographs. PVE leads to incorrect Region Of Interest (ROI) delineation and poor accuracy of quantitative and semi-quantitative image analysis. Despite current partial volume correction (PVC) methods where only ROI values are corrected, aim of this work was the implementation of a PVC approach of the full set of images based on the combination of PET with anatomical information obtained using computerized tomography (CT) or Magnetic Resonance (MR) images. The goals of the method were to increase the accuracy of ROI values and also to improve the spatial resolution of PET images for brain as well as whole body studies. Methods: PVC was performed by calculating the wavelet transform (WT) of PET-CT or PET-MR co-registered images. Wavelet decomposition was implemented using the a trous algorithm and higher resolution details from the corresponding anatomical image were inserted in the WT of the PET image after application of a linear functional. The performance of the proposed correction method was evaluated by calculating ROI mean value in simulated phantom PET images of structures with different grey levels intensity and size. Secondly the method was applied to true brain PET-MR and whole body PET-CT images. Results: ROI mean values of the uncorrected phantom images were about 20% lower with respect to the true value while for PVC images the difference was less than 5%. Global image quality in terms of spatial resolution was also tested by drawing line profiles across the corrected images. More precisely PVC brain PET-MR images showed a significant improvement in the delineation of small brain structures. Similar results were obtained using PET-CT images of the chest region. The difference in total number of counts in PVC images was about 0.6% with respect to the original image showing counts conservation. Conclusions: The implemented wavelet-based methodology for PVC showed a significant improvement in the measured ROI mean values, the PVC PET images spatial resolution was also significantly better.
- Society of Nuclear Medicine, Inc.