Attenuation artifacts and time-of-flight PET

Abstract

Attenuation artifacts in PET are seen when attenuation correction (AC) is not performed and when AC is performed but is based on an incorrect attenuation map. PET attenuation artifacts are generally more profound than the straightforward differences in photon attenuation through surrounding tissue from one source location to another. Effects such as the apparent radioactivity in gas pockets in the body or concave contours of the body surface, distortions, and a pronounced body contour are all in addition to the expected nonuniformities due to depth in the body. We have investigated the effects of time-of-flight (TOF) reconstruction on PET artifacts. Uniformity without AC (NAC) was investigated in a large tapering F-18-fillled phantom that increases from a small end to the size of a large patient. Throughout the length of the phantom, the NAC images with TOF were more uniform that the non-TOF images. A whole-body phantom with oval cross section was imaged with uniform F-18 background and a 5.5 cm air-filled sphere at one section, and a 5.5 cm hot (8? background) sphere at another section. In NAC images, the air-filled lesion was artificially hotter than background with non-TOF, but comparable to background with TOF. When this section was corrected with a uniform AC (as if the gas pocket had moved before transmission scan), the non-TOF image showed the sphere to be artificially hot (hotter than background) whereas the TOF images reduced the artifact. In the section with the hot sphere, the NAC distortions typical near the bladder (higher counts anterior and posterior to bladder; depleted areas lateral to bladder) were greatly diminished with TOF images. While AC is necessary for quantitation in PET, the use of TOF and NAC may be useful in providing more interpretable images in some situations were an attenuation map is impossible to obtain, or introduces errors of its own. TOF may also lessen some artifacts in corrected images.