SHORT COMMUNICATION MR-based attenuation correction for torso-PET/MR imaging: pitfalls in mapping MR to CT data Thomas Beyer & Markus Weigert & Harald H. Quick & Uwe Pietrzyk & Florian Vogt & Christoph Palm & Gerald Antoch & Stefan P. Müller & Andreas Bockisch Received: 23 October 2007 / Accepted: 8 January 2008 / Published online: 19 February 2008 # Springer-Verlag 2008 Abstract Purpose MR-based attenuation correction (AC) will become an integral part of combined PET/MR systems. Here, we propose a toolbox to validate MR-AC of clinical PET/MRI data sets. Methods Torso scans of ten patients were acquired on a combined PET/CT and on a 1.5-T MRI system. MR-based attenuation data were derived from the CT following MR– CT image co-registration and subsequent histogram match- ing. PET images were reconstructed after CT- (PET CT ) and MR-based AC (PET MRI ). Lesion-to-background (L/B) ratios were estimated on PET CT and PET MRI . Results MR–CT histogram matching leads to a mean voxel intensity difference in the CT- and MR-based attenuation images of 12% (max). Mean differences between PET MRI and PET CT were 19% (max). L/B ratios were similar except for the lung where local misregistration and intensity trans- formation leads to a biased PET MRI . Conclusion Our toolbox can be used to study pitfalls in MR-AC. We found that co-registration accuracy and pixel value transformation determine the accuracy of PET MRI . Keywords PET/MRI . PET/CT . Attenuation correction Introduction FDG-PET/CT imaging has emerged as a modality-of- choice for the diagnosis, staging and restaging of a variety of cancers [1]. While a single PET/CT examination provides complementary and intrinsically co-registered CT and PET images, the CT transmission images can be used for attenuation correction [2] as well, thus rendering lengthy PET transmission scanning with conventional rod or point sources obsolete. Thus, combined PET/CT imaging of the torso can be completed in 15 min or less. Recently, the combination of PET and MRI has been propagated as an alternative to existing dual-modality imag- ing, and the first images of patients have been presented in late 2006 [3]. This has raised many clinical expectations, but a realisation of PET/MR tomography beyond small- animal imaging remains technically challenging. In partic- ular, the lack of conventional or X-ray transmission sources mandates alternative approaches to attenuation correction of the complementary emission data. Attenuation correction in PET/CT, for example, is straightforward: attenuation coefficients at 511 keV are derived from the CT transmission images through a bi- linear segmentation-scaling approach that, in essence, separates bone from non-bone tissues followed by a subsequent scaling of the classified attenuation coefficients. While this works well for PET/CT imaging, it does not, however, for PET/MR since bone and air have similar Eur J Nucl Med Mol Imaging (2008) 35:1142–1146 DOI 10.1007/s00259-008-0734-0 T. Beyer (*) : M. Weigert : S. P. Müller : A. Bockisch Department of Nuclear Medicine, University Hospital, Hufelandstr 55, 45122 Essen, Germany e-mail: thomas.beyer@uni-due.de H. H. Quick : F. Vogt : G. Antoch Department of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Essen, Germany U. Pietrzyk : C. Palm Institute of Neuroscience and Biophysics, Research Centre Jülich, Jülich, Germany U. Pietrzyk Department of Physics, University of Wuppertal, Wuppertal, Germany