An Integrated Quality Index Metrology to Represent Multi-Feature Quality of Clinical Radiographic Chest Images
Presentations
SU-J-201-6 (Sunday, 7/10/2022) 4:00 PM - 5:00 PM [Eastern Time (GMT-4)]
Room 201
Purpose: Develop and implement a quality index (QI) to systematically quantify image quality of clinical radiographic chest images.
Methods: We developed a QI framework as an aggregate of 10 individual metrics of image quality extracted from clinical chest radiographs. The target value of each individual metric was determined from clinically accepted ranges reported in literature. We defined a normalized distance to target (NDT) as the difference between a measured image quality metric and its target value, normalized to an accepted tolerance range. The QI was defined using a root mean square combination of NDT values, bound between 0 and 1, where QI = 1 indicates optimal image quality under the specified constraints. To test our approach, we performed a retrospective analysis of clinical frontal chest radiographs from 9/1/2021-11/31/2021. We calculated the QI of each image to quantify the distribution of radiographic chest image quality throughout our healthcare system.
Results: In total, 11,956 images from 42 different imaging systems (40 direct digital, 2 computed radiography) were included in our analysis. The average QI across all images was 0.58 +/- 0.12 (range = [0.13, 0.84]). QI values stratified by unique imaging systems ranged from 0.40 to 0.67, demonstrating measurable differences in image quality between different systems. In our cohort, the metric measuring mediastinum detail was identified as the dominating effect driving QIs.
Conclusion: We developed a new approach to quantify the quality of chest radiographs in a clinical environment. Our data demonstrates the value of the QI in characterizing the image quality performance of different radiographic systems. Our QI framework enables systematic optimization of clinical chest radiography.
Keywords
Radiography, Chest Radiography
Taxonomy
IM- X-Ray: Quality Control and Image Quality Assessment
Contact Email