MRI-Based Attenuation Correction for PET/MRI Using a Morphologic Approach

• Main Authors: Fashun Tsai, 柴發順
• Other Authors: Jyh-Cheng Chen
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/33767457722141058331

碩士 === 國立陽明大學 === 生物醫學影像暨放射科學系暨研究所 === 99 === Puropose: PET / MRI can provide simultaneous anatomical and functional imaging, spectroscopy and molecular imaging information, which is the new platform for molecular imaging research. One of the challenges in PET/MRI is the derivation of an attenuation map to correct the PET image for attenuation. In this work, we develop a morphologic approach to obtain a pseudo CT image from corresponding MRI image for PET attenuation correction. Materials and Methods: We retrospectively selected 35 patients from the PACS system in our department, each patient received both MRI and PET/CT scans within five days. We combined MR T1-weighted image and T2-weighted image as an input MRI image data. Each MRI image was classified into three tissue classes (bone 1000 HU, soft tissue 35 HU, or air -1000 HU) by the following steps: After the input MRI image was classified into binary image, soft tissue and air were then differentiated using Otsu's method. Dilation was performed when the binary image was combined with a series of rank filters of different sizes. Bone region was segmented after subtraction the original binary MRI image from the dilated image. We made a quick MR image mask by region growing to correct exoskeleton. These pseudo-CT images were then used for attenuation correction, as the process would be performed in a PET/CT scanner. PET images reconstructed using the proposed MRI–based AC method (PETMRAC) were compared with the PET images that had been reconstructed using a CT-based AC method (PETCTAC) and images without AC (PETNOAC). Twelve ROIs were drawn on each image, including left and right sides of caudate nucleus, frontal lobe, temporal lobe, occipital lobe, parietal lobe and cerellum, and then two sample t-test was performed using SPM. Results: We assume the brain PET image with CT attenuation correction is the gold standard. PET image without attenuation correction is almost 86~90% less than PETCTAC image. The largest error of PETMRACamong all ROI is underestimated about 5.57%. SPM test shows that the PET image with MRI–based AC method was not significantly different from standard method’s results inside the brain area. Our MR-based attenuation correction could yield clinically acceptable errors compared with standard AC corrected images. Conclusion: Our proposed MRI-based attenuation correction approach is suitable for applications in brain PET imaging when the PET/MRI images are available. Considering the difficulties associated with transmission attenuation correction for PET/MRI scanner, MRI-based attenuation correction in brain PET would be the method of choice for hybrid PET/MRI scanners.