부산대학교 도서관

Clinical PET acquisitions of the heart suffer from artefacts and drops in image quality due to the poor spatial resolution of the PET system. Moreover, cardiac PET images are further degraded by the blur caused by the breathing and beating motions, thus hampering diagnosis and evaluation of myocardial pathologies. Anatomy-enhanced PET reconstruction, using a high-resolution CT, has proven useful in brain imaging. In cardiac datasets however, due to the motion artefacts, the application of any restoring technique on datasets affected by motion blur needs to be preceded by the validation of the proposed method on realistic static datasets. In this work, the validation is performed using static cardiac ex vivo datasets obtained from a number of sacrificed sheep, scanned on a clinical PET/CT scanner. The aim of this work is to assess the effectiveness of reconstructions of the acquired datasets with different CT-based anatomical priors, in comparison to reconstructions currently applied in clinical practise. The gold standard to which all reconstructions are compared consists of images of the same hearts scanned on a small-animal PET scanner, whose high spatial resolution allows for almost artefact-free images. Encouraging results were obtained so far, with improvements in volume delineation and uniformity of activity values when anatomical information was used. Fig 1 shows the gold standard image (left) compared to a regular clinical reconstruction (middle) and to a reconstruction using the high-resolution CT as anatomical information (right). Figure