Many recent reports are stating that image perception algorithms are very soon going to be better than humans, including medical imaging. What does this imply for radiology and radiologists? The routine use of Artificial Intelligence (AI) in medicine and in medical imaging will become reality, that´s a fact. This is true not only for radiology, but for all medical professions. The question is, how fast this may happen, for which indications and clinical scenarios, and how this will be implemented in our routine workflows. Indeed, implementation and use of AI in radiology is even beneficial and wanted, given the exponentially rising demand of diagnostic imaging and radiological procedures. But will AI replace our profession? Will we have to stop training radiologists? Radiology is always changing and has always been developing alongside new technologies. Therefore, AI will not replace our profession, but it will change our work and it will make us have to train even more. There are a number of reasons why diagnostic radiology has a bright future, implementing AI and ML. First, our workflows are getting more and more complex. From the decision to perform and indicate imaging, select the correct modality and procedure, to scheduling, scanning and acquisition to reading and post-processing, AI will help us to optimize patient care and throughput. Finally, the radiologist is the medical specialist to discuss and present our results to other clinicians and to our patients, including diagnostic reasoning and solving complex differential diagnoses. For the latter (communication and complex reasoning), AI will not be helpful, at least not for many years to come. Second, radiologist and any other doctor will always maintain responsibility for their medical practice, reports and treatments. Implementing AI will potentially increase safety and accuracy of certain procedures, giving the radiologist more time for communicating our results and for optimizing clinical care. However, AI will never be able to take over full legal responsibility. AI will deliver a “decision support”, let it be for an imaging diagnosis or, e.g., for selecting the appropriate oncologic treatment, but the decision-making will always be up to human doctors. Third, implementing AI will help us decrease waiting times, optimize scanning procedures, support decision-making and structuring our reports. Thus, it will help advancing the quality of our work, and consequently, demand for our work will even rise. The working profile of a radiologist will change in coming years. We will implement AI and ML, we will have to learn how to use and structure our data more efficiently, and how to correlate our findings better with other diagnostic entities such as lab tests, genetic tests or pathology. We may be happy not being forced to perform simple measurements on CT data or to report on many normal x-rays of a thorax. Radiology will be in the driving seat, delivering complex information, combining complex diagnostic and clinical information and data, and communicating those with our partner clinicians and patients. We will change the way we are being trained and the way we work – but certainly this future is nothing to be afraid of.

Acknowledgements

No acknowledgement found.

References

No reference found.