Quantitative attenuation correction for PET/CT using iterative reconstruction of low-dose dual-energy CT

We present the results of using iterative reconstruction of dual-energy CT (DECT) to perform accurate CT-based attenuation correction (CTAC) for PET emission images. Current methods, such as bilinear scaling, introduce quantitative errors in the PET emission image for bone, metallic implants, and contrast agents. DECT has had limited use in the past for quantitative CT imaging due to increased patient dose and high noise levels in the decoupled CT basis-material images. Reconstruction methods that model the acquisition physics impose a significant computational burden due to the large image matrix size (typically 512 × 512). For CTAC, however, three factors make DECT feasible: (1) a smaller matrix is needed for the transmission image, which reduces the noise per pixel, (2) a smaller matrix significantly accelerates an iterative CT reconstruction algorithm, (3) the monoenergetic transmission image at 511 keV is the sum of the two decoupled basis-material images. Initial results using a 128 × 128 matrix size for a test object comprised of air, soft tissue, dense bone, and a mixture of tissue and bone demonstrate a significant reduction of bias using DECT (from 20% to ∼0% for the tissue/bone mixture). FBP reconstructed images, however, have significant noise. Noise levels are reduced from ∼8% to ∼3% by the use of PWLS reconstruction.