METHOD AND SYSTEM OF MULTIVARIATE ANALYSIS ON NORMALIZED VOLUME-WISE DATA IN THE SINOGRAM DOMAIN FOR IMPROVED QUALITY IN POSITRON EMISSION TOMOGRAPHY STUDIES

摘要

A method and system are provided for improving the quality in positron emission tomography (PET) images. PET input data is masked using raw dynamic PET data (sinograms) as input for primary component analysis (PCA) that generates primary components which in turn are used to create a mask. This mask can be used to allow object pixel data to be extracted from the sinograms into masked sinograms where background pixels outside the reference object are set to zero. A volume-wise approach to PCA uses masked sinograms as input data. Pixel-wise noise pre-normalization may then be performed generating pre-normalized sinograms from the masked PET input data. PCA is then performed on the pre-normalized sinograms resulting in PCA sinograms recreated into PCA-modified sinograms by adding background pixel values of zero. These PCA-modified sinograms may optionally be scaled and may then be reconstructed into dynamic PET images with improved image quality.