Abstract
T5
Introduction: While brain perfusion single-photon emission computed tomography (SPECT) imaging has been widely used to measure regional cerebral perfusion and is commonly indicated for assessment of ischemic and hyperperfused areas with cerebrovascular diseases and differential diagnosis of various forms of dementia. It also provides functional information on the perfusion and metabolic status of the brain tissue, which cannot be obtained by structural neuroimaging, such as computed tomography and magnetic resonance imaging. SPECT image quality varies depending on gamma cameras and specific reconstruction and acquisition conditions even with similar radioactivity distribution. Physical parameters on brain SPECT are required for standardizing SPECT image quality, which should be measurable in brain phantom experiments and consistent with visual analysis. Such physical parameters allow for more objective evaluation of image quality in different facilities because visual analysis is easily affected to inter-facility variation, inter-observer variation, and observer’s experience. Therefore, this study aimed to evaluate the relationship between physical parameters of a brain phantom and visual analysis under various conditions in a multicenter trial study.
Methods: SPECT images of the Hoffman three-dimensional brain phantom were acquired from eight devices in five institutions. The phantom which simulates gray- and white-matter structures with 4:1 activity concentration was filled with 28 kBq/ml of 99mTc solution at the start of scanning. We obtained various data with different acquisition times under clinical reconstruction and acquisition conditions at each institution. Four physical parameters (%contrast, contrast noise ratio (CNR), asymmetry index (AI), sharpness index (SI)) were selected and measured with the phantom. Seven observers in nuclear medicine blindly evaluated all image series and scored 1–3 for four check points about contrast, image noise, symmetry, sharpness. The average score for all observers was calculated and evaluated based on the corresponding physical parameters, respectively.
Results: CNR showed good agreement with visual analysis of the contrast and image noise, both of which were significantly different between the group of "<2" and the group of "≥2 and <3". AI also became lower as the visual analysis score of symmetry increased, and both groups of "≥2 and <3" and "3" significantly lower from the group of "<2". Conversely, no relationship with visual analysis was found for %contrast and SI.
Conclusions: We clarified the relationship between physical parameters and visual analysis of a brain phantom in a multicenter trial study. CNR and AI showed good agreement with visual analysis, indicating their usefulness as physical parameters. Our findings of this study will help the quality of brain perfusion SPECT studies such as standardizing of the image quality, optimizing parameters for gamma camera upgrades, or obtaining reliable data in multicenter studies.