There’s an unfounded rumor floating around that Artificial Intelligence (AI) may one day replace radiologists. As a specialty field in medicine, the field is based in imaging, which allows the diagnosis and treatment of injuries and diseases. Today’s imaging techniques include X-rays, computed tomography (CT), positron emission tomography (PET), ultrasound (US) and magnetic resonance imaging (MRI). Treatment methods include radiation (beam or implanted seeds), nuclear medicine, and minimal to non invasive image guided treatments. Thus, radiology is a broad field filled with career opportunities and growth.
However, the idea that AI could somehow make human professionals obsolete is already stressing medical students who are considering a career in radiology. According to a 2021 survey, many medical students who are interested in radiology are rethinking that path.[i] They are worried that there won’t be jobs for them. I want to explain why I believe such anxiety is not premature, but totally groundless.
Where does AI come in?
Radiologists who interpret medical images are called readers. They are trained to makes sense of the specific anatomy and biology of what they are looking at, and to identify abnormalities. In a sense, this is similar to what you are doing right now as you read these words. You have been trained to recognize individual letters and punctuation marks, to make sense of how they are grouped into words and sentences, and to identify abnormalities like the two words you just read that stand out because they are in bold. Where does AI come into this?
Currently, AI is playing an increasing role to help readers. in interpreting medical imaging such as MRI and CT scans. AI and its components (Machine Learning, Deep Learning) is a work actively in progress. Thousands of medical images that are banked in digital case records are available to “train” sophisticated predictive algorithms to distinguish normal from abnormal tissues. The performance of these models is continually tested against human performance. To date, AI’s accuracy is comparable to that of expert readers—and in some studies, outperforms its human counterparts. Such outcomes are very promising in two important areas:
a) Reduced expenses of time and cost by accelerating accurate detection and diagnosis for busy radiologists, and
b) Better and faster results for patients.
As radiologists, we have an endless desire for efficient, consistent and highly accurate interpretation of clinical imaging. This is always in the best interest of our patients. It is crucial that radiologic readers around the globe be able to attain excellence in image interpretation, whether they are residents or experienced authorities. AI makes such consistency possible. AI is not intended to replace the human
brain, but rather to be integrated into the workflow of busy radiologists. In my own experience, as the vision is quickly becoming reality, it is a radiological dream come true, not a nightmare to be feared.
Of course, there are potential downsides to integrating AI in all levels of healthcare delivery, and I trust that medical ethicists are involved in the growth process of this young science. However, as a practicing clinician, my focus is pragmatic: I use AI tools in our Center’s mpMRI for prostate cancer detection, diagnosis, and MRI-guided focal laser ablation. In my experience, AI enhances the very high standards in which we take pride.
AI as partner in imaging
As Karolinska Institute’s Professor Lars Egevad expressed: “The idea is not for AI to replace human experts, but rather to function as a safety net to avoid pathologists missing cancer cases and to help in standardising the assessments. AI can also be an option in those parts of the world that today completely lack pathology expertise.”[ii] Most importantly, it is my personal belief that while the lightning quick predictive power of AI, based on access to hundreds or thousands of training datasets, can outpace the mental “library” that an individual radiologist has formed during his/her education and experience, it will never replace human judgment. AI is our partner in reading, but patients need to know that radiologists can override a software interpretation. Humans have the final say. Also, I have yet to find an AI model that has been trained on empathy, compassion, and that intangible, mysterious human capability we call intuition. If I feel it necessary in a patient’s case, I will trust my own conclusion more than the machine’s. I believe my colleagues would always do the same.
I don’t worry that AI is putting my job on the line. In fact, it has already directly or indirectly enhanced our patients’ outcomes. It is my hope that medical students who are considering radiology will take my word that properly engineered, trained and tested AI delivers extremely helpful information and insight. They should also have faith that the value of the heart and spirit they would bring to their patients is irreplaceable by any algorithm. Imaging in medicine is booming, and we need all the committed, compassionate radiologists we can get.
NOTE: This content is solely for purposes of information and does not substitute for diagnostic or medical advice. Talk to your doctor if you are experiencing pelvic pain, or have any other health concerns or questions of a personal medical nature.
[i] Reeder K, Lee H. Impact of artificial intelligence on US medical students’ choice of radiology. Clin Imaging. 2022 Jan;81:67-71. doi: 10.1016/j.clinimag.2021.09.018.
[ii] “Study shows AI systems can accurately identify and grade prostate cancer.” NewsMedical, Jan. 13, 2022. https://www.news-medical.net/news/20220113/Study-shows-AI-systems-can-accurately-identify-and-grade prostate-cancer.aspx