Transmission-less brain TOF PET imaging using MLACF

Recent theoretical investigations concluded that both activity and attenuation distributions can be obtained from PET emission TOF data alone up to the knowledge of the sinogram scaling parameter. Attenuation is recovered in the form of attenuation factors. This allows for development of a fast iterative algorithm, referred to as ML-ACF, where iterative reconstruction is performed with respect to the activity image. Attenuation is a by-product of the estimation of activity. In general, the estimation of attenuation from PET data alone is not currently necessary due to the diagnostic value of PET-CT scanning. Nevertheless, it is desirable to limit patient CT doses. CT-less PET brain imaging is currently being pursued by manufacturers. This transmission-less brain imaging uses the assumption of a uniform attenuation map. The boundary can be derived from the processing of the activity image and reconstructed without attenuation correction. In this paper we enrich this approach by exploiting this uniform attenuation map as a starting condition for MLACF iterations. The scatter correction and the proper scaling of the activity image are based on this information. The head, especially the slices close to the top of the head, is relatively small compared to account the current scanner's time resolution. Therefore it is not obvious a priori that TOF information can be exploited for brain imaging. In this paper we demonstrate that the brain image attenuation can be reconstructed by exploiting just two TOF bins, with the TOF bin size currently used on the Siemens mCT scanner. Preliminary investigations showed that the head activity image will be improved by MLACF compared to the standard CT-less processing based on a uniform attenuation.