Relationship between physical parameters and visual analysis for assessment of image quality: a multi-center and multi-vendor phantom study in brain SPECT.

Objectives: Brain perfusion single-photon emission computed tomography (SPECT) image quality varies depending on SPECT systems. This study aimed to evaluate the relationship between physical parameters and visual analysis for assessment of the brain SPECT image quality. We conducted our phantom study under various conditions in a multi-center and multi-vendor study.

Methods: SPECT images of the brain phantom were acquired from eight devices in five institutions. The phantom 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 (percent contrast, contrast noise ratio (CNR), asymmetry index (AI), and sharpness index (SI)) were measured with the phantom. Seven observers blindly evaluated all image series and scored them on a scale of 1-3 using four checkpoints: contrast, image noise, symmetry, and sharpness. The average score for all observers was calculated.

Results: CNR increased with increasing visual analysis scores for contrast and image noise, both of which were significantly different between the group with scores "<2" and the group with scores "≥2 and <3". AI decreased as the visual analysis score for symmetry increased, and the AI of both groups with scores "≥2 and <3" and "3" were significantly lower than that of the group with scores "<2". Conversely, no relationship with visual analysis was found for percent contrast and SI.

Conclusion: We clarified the relationship between physical parameters and visual analysis of a brain phantom in a multi-center and multi-vendor study. CNR and AI showed agreement with visual analysis.