Artificial intelligence improves radiologist workflow and assessment of image quality accuracy

Increasing imaging volumes and workflow pressures make consistent image quality assessment (IQA) challenging for radiologists in routine clinical practice. Artificial intelligence (AI) systems have recently been introduced to support quality control tasks, yet real-world evidence on their impact on radiologist workflow and diagnostic performance remains limited. Therefore, it is of interest to assess the effect of AI-IQA module of PACS/worklist on radiological workflow and accuracy in identifying suboptimal CT/MRI tests. Workflow (IQA time, TTFR and TAT) and IQA performance (sensitivity, specificity, kappa) metrics were evaluated in two groups (before and after 8 weeks of AI integration) and compared to an expert reference standard. AI integration lowered median IQA task time/examination and improved report turnaround time and improved radiologist sensitivity to suboptimal IQA without affecting specificity. Thus, we show AI-assisted IQA as a scalable tool to enhance the quality control and operation performance in the routine clinical practice.