A Radiographic Image Quality Framework for Diagnostic Accuracy Optimization

Thesis Abstract

This study addresses a persistent gap in radiographic practice where image quality variability undermines diagnostic accuracy across clinical workflows, particularly in conventional and digital radiography settings, leading to inconsistent patient outcomes and increased repeat-rate costs. The aim is to develop and validate a comprehensive Radiographic Image Quality Framework (RIQF) that integrates physical image quality metrics, radiographic technique factors, and interpretive performance to optimize diagnostic accuracy. Specific objectives include (1) identifying core image quality parameters (spatial resolution, contrast-detail, noise, artifact prevalence) and their relationship with diagnostic confidence; (2) examining how exposure indices, receptor type, and reconstruction algorithms influence perceived image quality and diagnostic decision-making; (3) formulating a theoretically grounded framework linking image quality constructs to diagnostic outcomes using the Technology Acceptance Model and the Information Processing Theory; (4) validating the framework through empirical assessment across multiple clinical departments with varying patient volumes; and (5) deriving practical guidelines for protocol optimization and training to reduce nondiagnostic imaging and unnecessary repeats. The methodological design adopts a mixed-methods approach, converging sequentially from quantitative measurement to qualitative interpretation. The population comprises radiographic examinations from three tertiary hospitals, with a target sample of 1,200 adult chest, abdomen, and extremity radiographs collected over six months, stratified by modality (CR, DR) and receptor size. Quantitative data will be gathered via standardized image quality scoring by a panel of five blinded radiologists using a validated 5-point Likert scale for each image, coupled with objective metrics captured automatically Modulation Transfer Function (MTF), Noise Power Spectrum (NPS), Signal-to-Noise Ratio (SNR), Detective Quantum Efficiency (DQE), exposure indices (KAP, E/I), and processed image reconstruction parameters. Qualitative data will be obtained through semi-structured interviews with ten radiologists and five radiographers to explore interpretation challenges, workflow constraints, and perceived impacts of image quality on diagnostic decisions. The instruments will be piloted in a preliminary cohort of 100 images to ensure reliability. Data analysis will proceed in two integrated strands. Quantitative analysis will employ multilevel linear modeling to assess associations between objective image quality metrics, exposure parameters, and diagnostic accuracy (ground-truth reference established through consensus reporting and, where available, follow-up imaging). Regression analyses will examine the predictive power of image quality indicators on diagnostic confidence and correctness, while ANOVA will compare performance across receptor types and modalities. Structural equation modeling (SEM) will test the hypothesized RI QF relationships, incorporating latent variables for image quality, diagnostic accuracy, and interpretive performance, and will evaluate mediation effects of radiologist experience and workflow efficiency. The qualitative data will be analyzed using thematic analysis to identify recurring themes related to the interpretation of image quality signals, cognitive load, and training needs, with coding conducted by two independent researchers and triangulated with quantitative findings. Expected findings include (a) robust correlations between objective image quality metrics (MTF, SNR, DQE) and diagnostic accuracy, moderated by modality and receptor type; (b) identification of critical thresholds for exposure indices beyond which diagnostic confidence deteriorates; (c) a validated RIQF model linking technical image quality dimensions to interpretive outcomes, and (d) practical guidelines for protocol optimization, including exposure settings, acquisition geometry, and post-processing workflows, tailored to department-specific patient populations. The study will also elucidate how radiographer training and radiologist feedback loops influence adherence to quality standards, informing targeted education interventions. The contribution to knowledge lies in establishing a theoretically informed, empirically validated framework that bridges radiographic physics, imaging workflow, and clinical interpretation to optimize diagnostic accuracy. It advances measurement standards by integrating objective physical metrics with perceptual and decision-making outcomes, and it provides a transferable model adaptable to diverse radiology services, across countries with varying resource levels. The recommendations will inform policy development for QA programs, revising image quality scoring systems, and guiding continuing professional development. The study concludes that implementing the RI QF with continuous monitoring of both objective metrics and interpretive performance can reduce nondiagnostic repeats, enhance diagnostic confidence, and ultimately improve patient care, with a structured roadmap for technology adoption, training, and clinical governance.