Performance evaluation of a positron tomograph designed for brain imaging

Abstract

The NeuroECAT, a multiplane positron tomograph for imaging the brain, was characterized in terms of both quantitative performance and image quality. The tomograph has four modes of operation, defined by the placement of interplane septa and shadow shields. Each mode was fully characterized by measurement of image resolution, axial resolution, resolution uniformity, scatter, accidentals, and deadtime. Each measurement was performed with scattering media simulating the human head, and resolutions were obtained from images processed with reconstruction techniques actually used in patient imaging. The results for the most frequent mode of operation are: image resolution 9.8 +/- 0.2 mm (FWHM), axial resolution 12.4 +/- 0.4 mm, and scatter 8.1% +/- 0.6. At a count rate of 10,000 cps per image plane, accidentals are 9% and the deadtime 3%. Accidentals are measured and subtracted in hardware, and corrections for deadtime loss are calculated from the on-line measurement of triple-coincidence events. Scatter is estimated from the scan data and subtracted in software. Image quality is demonstrated by phantom studies and by the patient images obtained with [F-18] fluorodeoxyglucose and carbon-11 monoxide. The FDG images show clear delineation of the convolutions of the cortical ribbon, internal gray nuclei, internal and external capsules, and other substructures of the brain. The carbon monoxide images, in addition to visualizing the large vessels, clearly show the blood volumes of the cortex, the Sylvian fissure, and the circle of Willis.