AI in projectional radiography

This chapter delves into the current and future transformative potential of AI-assisted optimisation in projectional radiography, namely plain X-ray imaging and mammography, exploring opportunities for enhancing every stage of the patient journey. Beginning with patient checks and planning, AI algorithms can streamline patient scheduling and support personalised care. During image data acquisition, AI can assist protocol selection and automate image acquisition tasks. AI techniques can also enhance image quality via radiation scatter reduction, smoothing and artifact correction, affording dose saving by mitigating the need for repeat projections. The chapter further explores AI’s role in data analysis, where machine learning models can probe image features to identify abnormalities and enhance workstreams via stratification of findings. In image evaluation and reporting, AI tools can automate report generation, highlight critical findings and support radiographers in making informed decisions as a decision support tool. Furthermore, the chapter addresses data quality and quality assurance, emphasising how AI can facilitate automated audit and remote monitoring ensuring quality, consistency and reliability. Finally, the chapter discusses the importance of not only optimal design of user interfaces but also customised radiographer training to be able to seamlessly work with new technologies like AI for the benefits of the patients. Case studies are presented to illustrate practical applications, such as automated abnormality detection in chest radiographs and spine imaging, image acquisition and image quality enhancements and AI-supported workflow enhancements. Challenges and limitations are contextualised in terms of data quality; integration into clinical workflows and the role of the radiographer in terms of accountability and ethical implementation are explored. The chapter aims to equip the reader with the baseline knowledge to be able to further interrogate advanced technologies and AI in projectional imaging, to shape the development of clinically useful tools for the ultimate benefit of the patient. The chapter concludes with a summary and final message lauding the adaptability of diagnostic radiographers and a call to not only embrace but also lead the change of AI-enabled diagnostic radiography.