Article

How Artificial Intelligence is Changing the Face of Radiology

Renowned theoretical physicist, Albert Einstein1, is credited with saying, “The world as we have created it is a process of our thinking. It cannot be changed without changing our thinking.” Artificial intelligence (AI) – in essence, a new way of thinking – is creating a game-changing shift in medical imaging. AI can help radiologists in speeding patient triage, reducing unconscious bias, and assessing overall patient health.

Current day demands of radiology 

Radiology, as a medical field, is flourishing. A study of trends in medical imaging reported by JAMA (The Journal of the American Medical Association) in 2019 indicated an annual percentage increase in computed tomography (CT) imaging examinations of 3.7% among adults in 2013-2016. Magnetic resonance imaging (MRI) exams for adults rose 1.3% from 2007-2016.2 The World Health Organization describes medical imaging as “crucial in a variety of medical settings [sic] and at all major levels of health care. In public health and preventive medicine as well as in both curative and palliative care, effective decisions depend on correct diagnoses.”3 

The face of the individual radiologist, however, is likely to show signs of fatigue and worry. RSNA News reported on a 2016 study by the American Medical Group Association finding that productivity of radiologists increased by 3.5% in 2015 and another 3.4% in 2016.4 In the article, Myriam Hunink, MD, Ph.D., professor of radiology at Erasmus University Medical Center, Rotterdam, the Netherlands, and adjunct professor of health policy at Harvard University’s T.H. Chan School of Public Health in Boston said, “Statistics show that, 10 years ago, we interpreted one image every 20 seconds; now we do one every three seconds.”4

GEHC_radiologist-salary-guide_750x120.png

It is a job that entails great precision and focus. Crucial decision points call for solid, experience-based judgments. There is tremendous pressure for a fast turnaround. Patients anxiously await answers. Referring physicians are eager to get treatment for those patients underway.

AI is emerging as the competency that complements the increasing role of radiology in diagnostic and interventional medicine while reducing stressors for the radiologist. 

How AI is helping to meet the demands of radiology 

AI is the theory and development of computer systems able to perform tasks that normally require human intelligence. Medical imaging generates large quantities of complex data. In a conventional radiology setting, evaluating those vast datasets and extracting meaningful information is a time-consuming process.

AI can provide efficiencies in administrative, scheduling, and billing workflows. However, the benefit to the radiologist, referring physician, and in due course to the patient comes in forms such as:

  • Speeding triage – According to the Centers for Disease Control and Prevention, hospital emergency departments in the United States saw 145.6 million patients in 2016. Of those visits, 39% were seen in less than 15 minutes, and 8.7% resulted in admission.5 With layers of mathematical equations and millions of connections and parameters that get trained and strengthened based on the desired output, deep learning can help to direct the radiologist to the patient most in need efficiently. For example, a 2019 Radiology Business article described a deep learning model for classifying pediatric elbow injuries as acute or nonacute. Jesse Rayan, MD, an imaging fellow at Massachusetts General Hospital said, "the model's greatest potential to improve patient care lies in ER triage, where studies that need more attention can be prioritized for quicker turnaround."6
  • Reducing errors – Unconscious bias (a human tendency to look for the expected), combined with work overload or inexperience can contribute to the radiologist missing something important in an exam. AI guides the reader in zeroing in on abnormalities that might otherwise be overlooked.
  • Predicting patient risk – A team from Massachusetts General Hospital and MIT’s Computer Science and Artificial Intelligence Laboratory developed a deep learning model to identify patterns in mammography images to predict future risk of cancer.6 The information supplied by this precise risk assessment can aid the physician in tailoring screening and prevention strategies.
  • Assessing overall health – A 2017 commentary in The Medical Futurist reviewed a study conducted at the University of Adelaide using deep learning algorithms to predict 5-year life expectancy. The AI system analyzed more than 16,000 image features indicative of disease, with a 69% accuracy rate, comparable to the rate for human diagnosticians.7 The goal of these researchers is to refine such an algorithm to serve as a measure of general health, rather than identification of a specific disease.
  • Clinical decision support – Possibly the most significant change related to AI in radiology is also the least visible. AI supports confidence in diagnosis without degrading turnaround time. Jessica Kent summarized this in a 2018 Health IT Analytics article, "AI and machine learning have demonstrated great potential in supplementing and verifying the work of clinicians, particularly in the complex field of imaging analytics."8

The face of radiology is changing, and AI is becoming one of its distinguishing features. It is no longer a fear of whether AI will replace radiologists. It is more a matter of radiologists who use AI replacing those who do not. 

References:

  1. Albert Einstein Facts. Biography. https://www.biography.com/scientist/albert-einstein August 27, 2019.
  2. Trends in Use of Medical Imaging in US Health Care Systems and in Ontario, Canada, 2000-2016T. JAMA Network. https://jamanetwork.com/journals/jama/article-abstract/2749213 January 9, 2020.
  3. Diagnostic Imaging. World Health Organization. https://www.who.int/diagnostic_imaging/en/ August 27, 2019.
  4. Radiology Salaries Increase, but so Do Workload and Burnout. RSNA News https://www.rsna.org/en/news/2017/october/radiology-salary-survey January 9, 2020.
  5. Emergency Department Visits. Centers for Disease Control and Prevention. https://www.cdc.gov/nchs/fastats/emergency-department.htm August 27, 2019.
  6. What Has Artificial Intelligence Done for Radiology Lately? Radiology Business. https://www.radiologybusiness.com/topics/ai-machine-learning/what-has-artificial-intelligence-done-radiology-lately August 27, 2019.
  7. The Future of Radiology and Artificial Intelligence. The Medical Futurist. https://medicalfuturist.com/the-future-of-radiology-and-ai August 27, 2019.
  8. How Artificial Intelligence is Changing Radiology, Pathology. Health IT Analytics Xtelligent Healthcare Media. https://healthitanalytics.com/news/how-artificial-intelligence-is-changing-radiology-pathology August 27, 2019.