AIM in Interventional Radiology

Abstract

Artificial Intelligence applications have recently demonstrated high diagnostic accuracy and increased workflow efficiency in radiology. Machine learning models, particularly deep learning, can perform complex tasks in medical imaging, especially when they are trained with a large amount of high-quality data. For effective realization of the potential of artificial intelligence in interventional radiology, some unique challenges involving data storage, interoperability, adoption of standards, and conflict of interest between physicians and developers need to be addressed. With immense innovation and technological breakthrough, the scope of interventional radiology is continuously increasing in both width and breadth, and so are the opportunities of AI to revolutionize the sector. Artificial intelligence can complement the efforts of the interventional radiologist through decision support, triaging and screening of patients, prevention of error, procedural and periprocedural support, patient monitoring, prognostication of diseases, outcome prediction, image acquisition, image processing, etc. In conjunction with augmented reality systems, it can also help in improving procedural skills of interventional radiology residents and fellows through superior simulation training. Artificial intelligence has a tremendous potential to boost the productivity of radiologists. However, they are unlikely to replace them as there are significant apprehensions regarding the legal accountability, transparency, fairness, equality, bias, or potential misuse by an artificial intelligence system which prevent any independent action or clinical application without the oversight of an expert radiologist.