Artificial Intelligence and Radiology: Collaboration Is Key.
J Am Coll Radiol. 2018 May;15(5):781-783. doi: 10.1016/j.jacr.2017.12.037. Epub 2018 Feb 2. Yi PH1, Hui FK2, Ting DSW3.
Previous article in issueNext article in issue Artificial intelligence (AI) and deep learning have been met with great interest by both the medical and nonmedical community 1, 2. Within medicine, deep learning has demonstrated abilities comparable to those of humans for the diagnosis of diabetic retinopathy , malignant melanoma , and tuberculosis . A crossroads between the tech world and medicine has emerged, with tech giants like Google (Menlo Park, California) and IBM (Armonk, New York, New York) working to develop health care AI programs .
In a recent editorial , Herbert Kressel, MD, wrote that the introduction of AI to the field of radiology “is a disquieting thought to many radiologists … [however,] all of us working in the health care arena … must come to terms with the benefits and challenges of enhanced artificial intelligence.” Although new investigations on the potential application of AI in radiology are published regularly, little focus has been made toward navigating this new intersection between the tech world and AI and radiology.
To better understand the landscape of this new crossroads between AI and radiology, we created a word cloud of the titles of the top 25 nonscientific news articles from a Google search for “artificial intelligence radiology” (Fig. 1). A word cloud visually represents a collection of words with the relative frequency of a word reflected by its size. From this word cloud, we identified four themes: (1) radiologists, (2) AI and machines, (3) optimism and ambition, and (4) skepticism and realism. In reflecting on these themes, we propose that not only can the seemingly conflicting aims of AI and radiology be reconciled, but they can be harmonized to create a mutually beneficial and synergistic effect for the betterment of radiology and humanity as a whole.