In positron emission tomography (PET), the correction of the AC of 511keV photons on their way through the patients’ body plays a key role in quantifying the tracer-uptake. However, the signal intensity in MRI is not related to linear attenuation coefficients (LACs). Thus, several approaches have been proposed to overcome this drawback. So far, the segmentation method is the most robust and straight-forward one for whole body imaging and has therefore been implemented into the software of most vendors [1]. Based on a T1-weighted sequence, it separates the body into different tissue types (e.g. background, lung, fat, soft tissue) and assigns every tissue type a predefined LAC to compute an attenuation map (µ-map). However, bone is routinely not implemented which can lead to underestimations of SUV in regions with a high amount of bony tissue [2]. Also, only one single LAC is defined for the lung tissue. Thus, the aim of this study was to analyze the accuracy of SUV quantification of the segmentation method in different regions of the lung in direct comparison to the CT-based AC and to measure the impact of regional differences of lung density as well as bone tissue.21 Patients (9 female, mean age 56.6±13.8 y) with examinations of at least substantial parts of the lung were examined in a PET/CT and subsequently in a fully integrated PET/MR (Biograph mMR, Siemens Healthcare). Acquired PET data from PET/MR were reconstructed using four different µ-maps: i) a CT-based µ-map (PET_CTAC , gold standard), ii) a CT-based µ-map in which the LAC of the lung tissue was replaced by the lung LAC from the MR-based segmentation method (PET­_CTAC_MRLUNG); iii) based on ii), the LAC of bone structures was additionally replaced with the LAC from the MR-based segmentation method (PET_CTAC_MRLUNG_NOBONE); iv) the μ-map from the vendor-provided MR-based segmentation method (PET_MRAC). µ-maps were modified using MATLAB. T2-weighted images as well as MR- and CT-based µ-maps from each patient were rigidly registered to the corresponding PET data from PET/MRI. 14 ROIs with a diameter of 1cm each were placed in physiological lung tissue of each patient using PMOD. For each lung, one ROI was set in the apex, three were set at the level of the hilum (anterior, middle, posterior) and three were set in the basal lung. The relative difference of SUVmean between the different PETs was defined as: (SUVmean (x) - SUVmean (y)) / SUVmean (y).Table 1 and figure 1 give overviews on the relative differences of PET reconstructions ii)-iv) in comparison to i). The replacement of lung tissue in PET_CTAC with LAC from MR (PET­_CTAC_MRLUNG) had no significant effect on the middle lung parts while a relative overestimation in the anterior parts and an underestimation in the posterior parts were observed. The additional replacement of bone tissue with soft tissue from MR (PET_CTAC_MRLUNG_NOBONE) had only very slight effects on the PET quantification in the anterior and middle parts while there was an additional increasing of SUV underestimation in the posterior regions. Compared to PET_CTAC_MRLUNG_NOBONE, a further relative underestimation of SUV in the middle and posterior regions were found in PET_MRAC. Figure 2 is an example of a subtraction picture of PET_CTAC and PET_MRAC.Both, the replacement of lung and bone tissue in the CT-based µ-map had strongest influence on the posterior lung parts leading to a relative underestimation. This suggests that the overall resulting deviation of SUV quantification between PET_MRAC and PET_CTAC is based on both, differences of lung density (e.g. gravitational dependency) and surrounding bone tissue. The additional difference between PET_MRAC and PET_CTAC_MRLUNG_NOBONE might be caused by the algorithm which has not replaced bone tissue completely in the CT-based µ-map because, according to the literature, the offset LAC was set to 0.11/cm while soft tissue LAC in PET/MR is set to 0.1/cm.Using the segmentation method for AC in PET/MR systems leads to deviations of SUV in physiological lung tissue, mostly affecting the posterior lung parts. Differences in lung density and the relative high amount of surrounding bone tissue in those regions (spine and costovertebral joints) seem to be strongly influencing factors. This has to be taken into account when comparing SUVs from PET/CT and PET/MR.