Abstract
1425
Objectives When reading whole body FDG-PET/CT studies, sagittal images are useful in evaluating the spine. On such images, we have occasionally seen isolated decreased FDG activity involving one or two lower thoracic vertebrae just above the level of diaphragmatic surface abutting the vertebral column. In such cases, further observation revealed coexistence of the so-called ‘Banana artifact’, i.e., attenuation artifact adjacent to the diaphragm (lung base and the dome of the liver and spleen) due to diaphragmatic motion. Based on this observation, we hypothesized that diaphragmatic motion can cause an artifactually decreased activity in the lower thoracic spine as well. The aim of this study was to formally evaluate 1) whether our hypothesis is correct and if so 2) its frequency.
Methods 154 PET/CT scans (81 males and 73 females, age 61±13 years) were analyzed. On the coronal CT images, the vertebra above where the hemi-diaphragmatic surfaces abut the vertebral column was defined as the target vertebra (TV). On sagittal PET images, FDG activity in TV was visually assessed. The SUV ratio of TV to the average of the SUV of 2 reference vertebrae, i.e., a vertebra 2 levels above and a vertebra 2 levels below the TV, was assessed (T/R). Then, the severity of banana artifact on coronal PET images was graded as 0 if no artifact, 1 if a thin photopenic stripe, and 2 if a thick banana-shaped photopenic artifact in the vicinity of diaphragm.
Results Visually decreased activity in TV was present in 0% (0/93), 3% (1/34) and 59% (16/27) of scans graded as 0, 1, and 2, respectively (Grade 0 and/or 1 vs. Grade 2, p < 0.0001). The T/R ratio was 0.99±0.14, 0.99±0.15 and 0.84±0.19 in Grade 0, 1, and 2 scans, respectively (p < 0.0001).
Conclusions Error of attenuation correction from respiratory motion can lead to underestimation of FDG activity in the lower thoracic spine; this should not be interpreted as possible marrow replacement process requiring further imaging if there is coexisting banana artifact.