Keywords:
Multicentre study, Observational, Not applicable, Quality assurance, Comparative studies, Digital radiography, Thorax, Professional issues, Artificial Intelligence, Chest
Authors:
M. Englmaier1, D. Sasse1, D. Pfeiffer1, M. Kotnik2, L. Lin2, H. J. Lamb2, J. Conradsen3, J. Fløtten3, N. Wieberneit4; 1Munich/DE, 2Leiden/NL, 3Herning/DK, 4Hamburg/DE
DOI:
10.26044/ecr2020/C-05601
Methods and materials
A total of 920 adult PA chest radiographs were assessed by six clinical experts from three European hospitals (two raters from each hospital) with regard to positioning quality. The images were selected from a publically available Chest X-ray dataset [9] to constitute a representative distribution of positioning quality.
The assessment was performed using a standardized rating tool. The evaluated quality aspects were based on the “European Guidelines on Quality Criteria for Diagnostic Radiographic Images” [10]:
(a) Field of view, in which each side of the image was rated individually, as illustrated in Fig. 1. The distance of the lung to the respective edge of the image was assessed and could be too narrow or too wide. The rating was given in the tool shown in Fig. 4. 200 images were rated for the cranial and abdominal aspect ("North" and "South"), respectively, and 100 images for both left and right side ("East" and "West").
(b) Axial rotation, assessed based on the symmetry of the clavicle heads with respect to the center of the spine, as illustrated in Fig. 2. Images could be rated as symmetrical or asymmetrical. 200 images were rated with respect to rotation symmetry.
(c) Inhalation state, which was evaluated based on the number of anterior ribs visible above the diaphragm, as illustrated in Fig. 3. Inhalation could be rated as full or incomplete. 120 images were rated for this quality aspect.
Images were rated as good, inadequate or unacceptable for each quality aspect. The results were then analyzed both per hospital and across hospitals. Fleiss Kappa values were calculated for all image quality aspects.