Premium
Correlation between physical measurements and observer evaluations of image quality in digital chest radiography
Author(s) -
Yalcin Asena,
Olgar Turan,
Sancak Tanzer,
Atac Gokce Kaan,
Akyar Serdar
Publication year - 2020
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1002/mp.14244
Subject(s) - image quality , imaging phantom , digital radiography , radiography , computed radiography , medicine , nuclear medicine , correlation , radiology , mathematics , artificial intelligence , computer science , image (mathematics) , geometry
Purpose The aim of this paper was to investigate the relationship between the physical and subjective (observer) image quality metrics in digital chest radiography. Methods Five digital radiography systems, four with indirect flat panel detector and one with storage phosphor‐based computed radiography system, were used in the study. The physical image quality assessments were carried out using effective detective quantum efficiency (eDQE) metric and subjective performance of the digital radiography systems was evaluated in terms of inverse image quality figure (IQF inv ) derived from the contrast‐detail (CD) diagrams using CDRAD 2.0 phantom and CDRAD phantom analyzer software. All measurements were performed for different tube voltages (70, 81, 90, 102, 110, and 125 kVp) and polymethyl methacrylate (PMMA) phantom thicknesses. An anthropomorphic chest phantom and visual grading analysis (VGA) technique based on European image quality criteria for chest radiography were used for clinical image quality evaluation. Results The Spearman correlation coefficients were calculated for the investigation of the correlation between physical image quality and clinical image quality. The results showed strong positive correlation between the physical and clinical image quality findings. The minimum correlation coefficient was 0.91 ( p < 0.011) for IQF inv vs VGA scores and 0.92 ( p < 0.009) for IeDQE vs VGA scores. Conclusions Our results confirm that clinical image quality can be predicted with both physical assessments and contrast‐detail detectability studies.