Parametric mapping of blood flow and arterial blood volume in dynamic O-15 water brain PET by cluster analysis

A method for simultaneous parametric mapping of both cerebral blood flow (CBF) and cerebral blood volume of artery (CBV_a) from single dynamic water image is presented. This method incorporates the 2-compartment model including arterial vascular volume (V_a), and cluster analysis for suppressing noise in generating parametric image. A dynamic O-15 water PET scan was performed, after a static O-15 CO gas scan. Arterial input function was obtained during the dynamic scan. Using the reconstructed dynamic image, both images of CBF and CBV_a were generated by the presented method. Both generated images of CBF and CBV_a were compared to those generated by conventional autoradiography (ARG) and CO image. The CBF image by this method showed the elimination of CBV_a component, and was comparable with that by ARG in terms of quantitative CBF value and noise. The CBV_a image showed the spatial distribution of activities of pure arterial blood volume only, quantitative value and suppressed noise, in contrast to the image of total blood volume from CO image including artery and vein together. The new method with cluster analysis could create the images with quantitative values of CBF and V_a, with improved signal-to-noise ratio. The image of CBV_a provided the intuitional information of arterial blood volume only This method could contribute to improved diagnosis of cerebrovascular disease using O-15 water PET.