Compton scatter and X-ray crosstalk and the use of very thin intercrystal septa in high-resolution PET detectors

Abstract

To improve spatial resolution, positron emission tomography (PET) systems are being developed with finer detector elements. Unfortunately, using a smaller crystal size increases intercrystal Compton scatter and X-ray escape crosstalk, causing positioning errors that can lead to degradation of image contrast. We investigated the use of extremely thin (<300 /spl mu/m) lead strips for passive shielding of this intercrystal crosstalk. Using annihilation gamma rays and small (2- and 3-mm wide) bismuth germanate (BGO) crystal detectors in coincidence, crosstalk studies were performed with either two small adjacent crystals [(one-dimensional) (1-D)] or one crystal inside a volume of BGO [(two-dimensional) (2-D)]. The fraction of Compton scattered events from one crystal into an adjacent one was reduced, on average, by a factor of 3.2 (2.2) in the 1-D experiment and by a factor of 3.0 (2.1) in 2-D one, with a 300 (150)-/spl mu/m-thick lead strip in between the crystals and a 300-700-keV energy window in both crystals. We could not measure a reduction in bismuth X-ray crosstalk with the use of lead septa due to the production of lead X-rays of similar energy. The full-width at half-maximum (FWHM) of the coincident point-spread function (CPSF) was not significantly different for the 1- and 2-D studies, with or without the different septa in place. However, the FWTM was roughly 20% smaller with the 300-/spl mu/m lead shielding in place. These results indicate that intercrystal crosstalk does not affect the positioning resolution at FWHM, but does affect the tails of the CPSF. Thus, without introducing any additional dead area, an insertion of very thin lead strips can reduce the extent of positioning errors. Reducing the intercrystal crosstalk in a high-resolution PET detector array could potentially improve tomographic image contrast in situations where intercrystal crosstalk plays a significant role in event mispositioning.