Investigating the use of non-attenuation corrected PET images for the attenuation correction of PET data in PET/MR systems

Abstract

MR-based attenuation correction (AC) in PET/MR imaging is currently evaluated by many groups. The aim of this study is to investigate the feasibility of using the non-attenuated PET images (PET-NAC) as a means for the AC of PET images in PET/MR systems. A water fillable torso phantom with hollow spheres was scanned on a GE Discovery-RX PET/CT scanner. The activity concentration in the spheres and background was 37.2 and 6.27 kBq/cc respectively. A PET scan was acquired in 3D using 2 FOV with 5min / FOV. CT-AC PET (PET-CTAC), PET-NAC, and CT images were generated. In addition the PET-NAC images were used to create an attenuation map through an iterative segmentation process in three steps. In the first step of the process, the phantom body contour was segmented from the PET-NAC image. Voxels inside the contour were then assigned a value of 0.094 cm−1 to represent the attenuation coefficient of soft tissue at 511 keV. This segmented attenuation map was then used to attenuate correct the raw PET data and the resulting PET images were used as the input to the second step of the process. The lung region was segmented in the second step and voxels were assigned a value of 0.024 cm−1 representing the attenuation coefficients of lung tissue at 511 keV. The updated attenuation map was then used for a second time to attenuate correct the raw PET data, and the resulting PET images were used as the input to the third step. The purpose of the third step is to delineate parts of the heart and liver from the lung contour since these parts were unavoidably included in the lung contour of the second step. These parts were then corrected by using a value of 0.094 cm−1 in the attenuation map. Finally the attenuation coefficients of the scanner couch were included based on the CT images to eliminate the impact of the couch on the accuracy of AC. This final attenuation map was then used to attenuate correct the raw PET data and resulted in a final PET image (PET-IAC). Visual inspection and SUV measurements of the spheres between PET-IAC and the PET-CTAC were performed to assess the accuracy of image quantification. Furthermore, our proposed approach was used on two patients with lung and liver tumors to assess its feasibility in clinical studies. The results show that there is very small difference between the PET-IAC and PET-CTAC from visual inspection. PET-IAC SUVs are on average equal to 94±3% compared to the SUVs from the PET-CTAC in phantom study. For the two patient studies, this value is 111±12%. In conclusion, the use of PET-NAC as a means for the AC of PET images is feasible in the clinic. Such approach can potentially be an alternative method of MR-based AC in PET/MR imaging.