Harmonizing quantitative imaging feature values in CT using image quality metrics as a basis

• DOI: 10.1117/12.3006885
• Publication Date: 2024-04-02T19:58:25Z
• Authors: Morgan Daly; John M. Hoffman; Andrew M. Hernandez; Ali Uneri; Jeffrey H. Siewer...; Paul E. Kinahan; Nicholas B. Bevins; Kalpana M. Kanal; David A. Zamora; Benjamin W. Maloney; J. Anthony Seibert; Mark P. Supanich; M. Mahesh; John M. Boone; Michael F. McNitt...
• Keywords: Feature (linguistics); Optical transfer function

This study is an initial investigation into methods to harmonize quantitative imaging (QI) feature values across CT scanners based on image quality metrics. To assess the impact of harmonization QI features, we: (1) scanned assessment phantom three over a wide range acquisition and reconstruction conditions; (2) from those scans, assessed for each scanner at condition; (3) these assessments, identified set parameters that yielded similar ("harmonized condition"); (4) second with texture (i.e., local variations in attenuation) under same (5) extracted features compared between non-harmonized harmonized conditions. Quantitative assessments provided contrast noise ratio (CNR) modulation transfer function frequency 50% (MTF f50) condition used. A conditions was similarity CNR MTF <i>f</i>₅₀. provide comparison, several sets were identified. From phantom, standard deviation (intensity mean variance, GLCM autocorrelation cluster tendency, GLDM high low gray level emphasis) systems decreased 72.8% 81.1% unharmonized groups (with exception intensity which showed little difference scanners). These results suggest selecting protocols produce metric different can reduce variance systems.