Chance und Risken von künstlicher Intelligenz und Machine Learning in der bildgebenden Diagnostik

Artificial intelligence (AI) is increasingly being employed in diagnostic imaging. This is a broad term encompassing computer programs capable of undertaking and solving intelligent tasks. The evolving complexity of AI architectures enables more demanding tasks, such as the recognition and quantification of radiological parameters to be overcome with more sophistication . Currently, in the majority of cases the assessment and description of such parameters are carried out manually and in a narrative form. This manual evaluation process is not only time-consuming but also susceptible to interrater and intrarater variability, as it is strongly influenced by the person doing the assessment and external factors. Using AI-algorithms, standardized and reproducible results can be generated as it can exactly evaluate information from imaging data down to the individual pixels, independent of external influences. A decisive advantage is that in contrast to manual assessment, AI has the capability to incorporate extensive background data into the evaluation, which leads to a further enhancement of the precision. Functioning as a supportive tool, AI can elevate the quality of X‑ray image assessment while concurrently alleviating the workload.