Contents I. Introduction
Accurate and patient-specific corrections for photon attenuation and scatter are necessary to improve the quantitative accuracy of reconstructed SPECT images. These corrections require information about the distribution of linear attenuation coefficients in the body, also called an attenuation map. In SPECT these maps can be obtained with a transmission scan, in which an external source of radioactivity is used. However, these attenuation maps often contain artifacts [1]. Alternatively, attenuation-maps can be obtained using the high-resolution anatomical information available from other imaging modalities, such as X-ray CT or MRI. In both of these cases the linear attenuation coefficient necessary for SPECT must be surmised from another quantity (i.e., the averaged linear attenuation coefficient for a spectrum of X-ray energies in CT or the proton density in MRI). In practice, this involves segmenting the MRI or CT image into known anatomical regions and assigning the appropriate attenuation coefficient to each region. An additional step may also be required to spatially co-register the SPECT image with that from the other imaging modality.
