| ABSTRACT |
Nonstandard PET radionuclides like 124 I, 76 Br, or 86 Y, emit prompt gamma rays together with the positrons which can be detected in coincidence with annihilation photons, creating spurious double coincidences. This causes a background in the PET images that reduces contrast and hampers quantitative accuracy. Existing correction methods are computationally intensive and since they depend on the size of the subject and on the accuracy of a number of complicated estimates, they may be valid for large source distributions only up to a certain size. To overcome these challenges, we have implemented a novel correction which is based on the measurement of positron+gamma triple coincidences.
In such events, only one of the possible linesofresponse (LOR) defined by the triple interaction is the correct one, while the others correspond to the same type of spurious background present in the double coincidences. We use triple coincidences to iteratively separate those LORs into two different datasets, one corresponding to the ‘signal’ and the other corresponding to the ‘spurious background’.
These datasets are then combined with the standard doublecoincidences within an iterative reconstruction algorithm to generate backgroundcorrected images with improved contrast. We evaluated the proposed methodology in a preclinical PET/CT scanner, modified to enable the acquisition of triple coincidences, using a contrast phantom filled with 124 I. Our methodology provides a significant improvement in image contrast (up to 23%) in hot lesions and an extra gain in sensitivity due to the inclusion of recovered triple coincidences which leads to images with up to 6.8% less noise.
|
