Quantitative Similarity Between Digitally Reconstructed and Digital Radiographs Across Computed Tomography Slice Thicknesses: A Phantom Study

Purpose: This study quantitatively assessed intensity-based similarity and structural consistency between digitally reconstructed radiographs (DRRs) generated from computed tomography (CT) datasets reconstructed with different slice thicknesses and corresponding digital radiography (DR) images.Methods: CT images of an anthropomorphic phantom were acquired with slice thicknesses of 1, 2, 3, and 5 mm, and DRRs were generated from each CT dataset.Resampling and rigid registration were performed before comparison with the corresponding DR images.Image similarity was evaluated using normalized cross-correlation (NCC), mutual information (MI), and the structural similarity index (SSIM).Structural consistency was assessed using the Edge Dice index.Differences among slice thickness conditions were evaluated using the Friedman test, and correlations between the Edge Dice index and intensity-based similarity metrics were assessed using Pearson correlation analysis.Results: NCC, MI, and SSIM values remained similarly high at slice thicknesses of 1 and 2 mm but generally decreased with increasing slice thickness.Edge Dice values were comparable among the 1-, 2-, and 3-mm conditions but decreased at 5 mm.The Edge Dice index was positively correlated with NCC (r=0.828),MI (r=0.859), and SSIM (r=0.885).Conclusions: DRRs generated from CT datasets showed high intensity-based similarity to the corresponding DR images across all evaluated slice thicknesses.Slice thickness affected both intensity-based similarity and structural consistency, with the highest and most consistent similarity observed at 1 and 2 mm and lower similarity observed at greater slice thicknesses.These findings provide quantitative reference data for future investigations of CT-derived DRR characteristics and their potential imaging applications.