A Framework for Optimizing Radiographic Dose with Image Quality Theory

Thesis Abstract

Radiographic imaging remains a balance between diagnostic image quality and patient radiation safety, with variability in dose often not commensurate with clinically meaningful improvements in diagnostic information. This study addresses the persistent gap between dose optimization and image quality optimization by proposing a framework that integrates dose management with image quality theory to guide parameter selection in clinical radiography. The aim is to develop and validate a practical framework that prescribes dose levels aligned with quantified image quality targets across common radiographic procedures. Specific objectives include (1) delineating an operationalized image quality index compatible with clinical workflow, (2) calibrating dose-image quality relationships for chest, abdomen, and extremity radiographs, (3) evaluating the framework’s capacity to reduce patient dose without compromising diagnostic confidence, and (4) generating a decision-support tool that integrates patient factors, technique selection, and imaging system characteristics. A mixed-methods design is employed. The methodological core combines a quantitative experimental component and a qualitative evaluative phase. The quantitative strand uses a cross-sectional sample of 420 radiographic examinations across three departments in a tertiary hospital, stratified by modality (digital radiography and computed radiography) and body region (chest, abdomen, extremities). Simulated phantoms and clinical cases provide standardized data for measuring image quality metrics, including contrast-to-noise ratio, modulation transfer function, and detectability indices, while recording entrance surface dose. Regression analyses and dose-optimization simulations will establish dose-image quality functions and identify procedure-specific target dose ranges. The qualitative strand engages 20 radiologists and radiographers through semi-structured interviews and a Delphi panel to refine the theoretical underpinnings of image quality targets, confirm clinical relevance, and evaluate the acceptability of the proposed decision-support tool. The study draws on established theories in radiography, notably the Image Quality Theory by Winston and the Optimization of Radiation Dose by the ALARA principle, supplemented by Bayesian decision theory to model uncertainty in image quality assessments and dose recommendations. Data collection instruments include calibrated dosimeters, standardized image quality phantoms, a digital imaging system QA/QA log, observer rating forms for visual grading analysis, and a structured interview guide. Validity and reliability are pursued through calibration of phantoms, inter-rater reliability checks (? statistics) for image quality assessments, test-retest reliability analyses, and pilot testing of instruments. Data analysis employs multivariate regression to quantify dose-image quality trade-offs, generalized linear modeling to account for non-linearities in the dose-response, and ANOVA to compare performance across modalities and body regions. Model specification will integrate dose metrics, image quality indices, and clinical decision thresholds to derive a practical framework with an accompanying algorithm. Thematic analysis will synthesize qualitative data to elucidate professional considerations and practical constraints, with triangulation to ensure corroboration across data sources. Expected findings include robust dose-image quality curves for each body region and modality, identification of safe-dose windows that preserve diagnostic accuracy within predefined image quality targets, and a validated decision-support prototype that recommends technique factors and exposure settings. The study is anticipated to demonstrate that targeted dose reductions of 15–25% are achievable for chest radiographs and 10–20% for abdominal and extremity studies without observable degradation in diagnostic confidence, when guided by the framework. The contribution to knowledge lies in operationalizing a theoretically grounded framework that links image quality theory with dose optimization in routine radiographic practice, providing a replicable methodology for other institutions and informing standard-setting bodies. The study will also advance methodological integration by combining quantitative dose-image quality modeling with qualitative insights from radiology professionals, thereby enhancing the framework’s transferability and adoption potential. The main conclusion is that a theoretically informed, empirically validated framework can optimize radiographic dose while maintaining clinically adequate image quality, with a practical decision-support tool facilitating adoption within busy radiology departments. Recommendations include integrating the framework into radiographer training curricula, embedding the decision-support tool within imaging workflow software, periodic revalidation of image quality targets as technology evolves, and expanding the framework to include dynamic, patient-specific factors such as age, body habitus, and comorbidity profiles. Further research is suggested to test longitudinal outcomes on diagnostic accuracy and patient safety across diverse clinical settings.