Whether it’s through use of electrocardiograms or computer-aided detection (CAD) technology in breast imaging, AI is in
health care facilities, and doctors and technicians have been using it for years.
“Going forward, it will become incumbent upon radiologists
to understand the tools and understand how to integrate them
into their practice,” Kahn said.
How Automation Is Being Used Today
Machine learning, a computer science discipline that is a subset of AI, has had a tremendous effect on the advancements in
radiology CAD, according to an article published in the American
Journal of Roentgenology ( AJR) that investigated the automated
analysis of abdominal C T. A recent advance in computer science
is the refinement of neural networks, a type of machine learning
classifier used to make decisions from data. This refinement,
known generically as deep learning but more specifically as convolutional neural networks, has shown dramatic improvements
in automated intelligence applications [ 4].
Initially drawing attention for impressive improvements in
speech recognition and natural image interpretation, deep learning is now being applied to medical images. According to the July
2016 AJR article, “Progress in Fully Automated Abdominal CT
Interpretation” by Ronald M. Summers, the results from AI have
been particularly promising for the reduction of false-positives.
Because of these deep learning techniques, Summers indicated that advances in abdominal C T automated image interpretation are occurring at a rapid pace. In the not too distant future,
these advances may enable fully automated image interpretation. Furthermore, he said similar advances may occur in other
body regions and with other imaging modalities.
“Risks and benefits are difficult to foresee but may include in-
creased pressures for commoditization, better reading efficiency,
fewer interpretive errors, and a more quantitative radiology re-
port,” Summers wrote. “The primary focus must ultimately be on
improved patient care.”
To perform fully automated abdominal CT image interpretation
at the level of a trained radiologist, Summers said the computer
must assess all the organs and detect all the abnormalities present
in the images. Although this is a seemingly daunting task for the
software developer, the numbers of organs and potential abnor-
malities are finite and can be addressed methodically.
Advances are likely in several areas pertaining to automated
abdominal CT image interpretation. These areas include machine learning, big data, automated report generation, multimo-dality image analysis, publicly available datasets and competitive
challenges, investigation of other organs and diseases, and new
applications. If automated interpretation is widely realized, there
will be effects on radiologists that will need to be considered.
Is There Danger Ahead?
Well, that depends on whom you ask. In an article published
in Information Age in October 2016, the author quoted theoreti-
Artificial Intelligence continued from p. 3 cal physicist Stephen Hawking warning that “AI could be the
greatest disaster in human history, unless humans learn to miti-
gate the risks posed.” In the same presentation, Hawking warned
that “machines could develop a will of their own” [ 5].
While cautious, others don’t view AI in such sinister terms. In
fact, Dreyer pointed out that certain media coverage can be
harmful to the industry’s messaging about AI.
“It is challenging when the lay press writes articles implying
that AI will overtake radiology,” Dreyer said. “Medical students
read these articles and decide not to study radiology. We need to
be more transparent about AI and do a better job in education.”
Radiologists may not be in danger of losing their jobs, but
their jobs likely will change.
According to Summers, “Autopilots for airplanes changed the
role of the pilot. Self-driving cars will change the role of the driv-
er. In both cases, the human is still ultimately responsible for the
safety of the passengers. Similarly, fully automated abdominal
CT image interpretation is likely to change the role of radiolo-
gists, but they will still be responsible for taking care of the pa-
tient and making the final diagnosis. Radiologists must be
vigilant to avoid placing too much trust in the computer.”
Like his peers, Kahn views the use of AI as a favorable shift for
the radiologists and patients but cautions that there are bad actors
out there. Not all machines and equipment will be created equally.
This raises the question about the FDA and how it will regulate AI.
Dreyer said regulatory issues may be an issue as the technology continues to evolve and the FDA is forced to keep up.
According to Forbes, in July 2016, the FDA issued guidance on
future medical innovation stating that it does not foresee placing
an undue regulatory burden on it. Forbes opinion contributor
John Graham said current law does not define the FDA’s powers
to regulate devices that depend on advances in AI and machine
learning as applied to health care [ 6].
As AI continues to advance and impact various industries including health care, it could improve medicine beyond today’s
capabilities. When asked how AI will change radiology and patient care, Kahn said, “No one knows—it’s too early to tell.”n
1. Executive Office of the President, National Science and Technology Council,
Committee on Technology. Preparing for the Future of Artificial
pdf. Published October 2016. Accessed March 23, 2017
2. Hernandez D. Artificial intelligence is now telling doctors how to treat you.
Wired. https://www.wired.com/2014/06/ai-health care/. Published June 2,
2014. Accessed March 23, 2017
3. American College of Radiology (ACR). ACR responses to request for information on artificial intelligence. https://www.acr.org/~/media/ACR/
acr_comments_ostp_rfiai_7222016.pdf?la=en. Published July 22, 2016.
Accessed March 23, 2017
4. Summers, RM. Progress in fully automated abdominal CT interpretation.
AJR 2016; 207:67–79
5. Ismail, N. AI: the greatest threat in human history? Information Age. http://
Published October 20, 2016. Accessed March 23, 2017
6. Graham, J. Artificial intelligence, machine learning and the FDA. Forbes.
intelligence-machine-learning-and-the-fda/#37b1cc901aa1 Published Aug.
19, 2016. Accessed March 24, 2017