Evaluating the relationship between objective and subjective image quality in computed tomography using detectability index and three-dimensional-printed phantoms

Achieving optimal image quality in computed tomography (CT) is essential for accurate diagnosis. This study explores the relationship between objective and subjective image quality, using the detectability index ($d^{\prime}$) as a quantitative performance metric. A 3D-printed anthropomorphic abdominal phantom was used to assess how variations in tube voltage, current, slice thickness, and reconstruction kernel affect image quality. Signal-to-noise ratio, contrast-to-noise ratio, and edge sharpness were evaluated alongside radiologist assessments. A strong correlation between $d^{\prime}$ and subjective quality confirmed $d^{\prime}$ as a reliable predictor of visual image perception. Multiple linear regression identified $d^{\prime}$, object size, and the peak spatial frequency of the noise power spectrum as independent predictors of perceived quality. These results support the integration of $d^{\prime}$ into routine CT image quality evaluation, helping to align technical image parameters with clinical expectations and improve protocol optimization.