Feature article

Five AI Imaging Findings That May Be Game Changers

For all the chatter about the artificial intelligence (AI) splash in the imaging pool, the reality is that AI has been an influential technology for some time. The term "AI" was coined in 1956 to designate a machine that exemplifies human intelligence,1 and today, that definition has not changed much. A 2016 Stanford University report defined AI as, “a science and set of computational technologies that are inspiredbut typically operate quite differentlyby the way people use their nervous system and bodies to learn, reason, and take action.”2 The takeaway for radiologists is that AI has expanded to create neural networks with applications that may be able to take over their daily mundane and repetitive tasks. And as of June 2018, there are 98 companies that are vested in producing AI solutions to medical imaging.3 According to Dr. Kurt Shoppe, a radiologist and Chair of the Reimbursement Committee at the American College of Radiology (ACR), AI is a tool to bring better insight, and even joy, to the practice of radiology.

1. Imaging warms to AI

Even though imaging applications are just emerging, early adoption will be rewarded. In five to 10 years, early adopters will realize significant efficiency and quality gains to achieve a competitive advantage.5 Most radiology leaders have moved beyond seeing AI as just a buzzword. In a 2017 study of imaging professionals6 including radiologists and chiefs of radiology, 84 percent viewed machine learning as important or extremely important. Only 16 percent of the cohort said there were no plans for their organization to adopt imaging machine learning. Of this segment, nearly half described their organization as "not forward-looking."

2. Radiologists will control the pace of change

The ACR has partnered with The Medical Image Computing and Computer Assisted Intervention Society and signed an agreement in May 2018 to advance the use of AI in the imaging sector. A compelling reason for this joint effort is to take charge of AI development and to promote standards for good radiology workflow.7 The two entities are also seeking to develop AI that is safe, effective, and has the maximum impact on improving patient care. An early adopter of AI imaging, a medical practice of 50 radiologists, Wake Radiology Medical Imaging in North Carolina, has trialed 53 different AI platforms. Since the implementation of their trial program, the practice has purchased five of the AI engines to work with its picture archiving and communication system.8

3. New AI techniques contribute to value-based delivery

Imaging AI is a critical part of the solution to the challenge of finite medical resources.9 With a projected shortage of radiologists10 and the increasing worldwide demand of an aging, sicker population, AI can create efficiencies critical for value-based delivery (VBD). VBD, or outcomes medicine, is the intersection between cost efficiency, quality outcomes, and patient centricity.11 The goal is to make society healthier while reducing healthcare spending.12 In the U.S., imaging AI is seen as a critical tool in lowering healthcare spending from the current high cost of 18 percent of the gross domestic product.13 Imaging is leading the way in the medical field and demonstrating how to deploy AI in a data-rich environment.14 For example, a long-standing reality in imaging is that about 70 percent of patients do not follow-up with additional imaging within six months of a radiologist’s recommendation. By deploying AI data mining and quality analytics, follow-up rates could improve, and that would also generate additional revenue. Also, early diagnostics would lower the incidence and expense of late-stage disease management.15

4. AI financing and modeling are aligned to transform imaging

Sigal Atzmon, the chief executive officer and president of Medix, a healthcare and scientific staffing company, terms data and AI as the "kings of wisdom." She cautioned that healthcare institutions that do not embrace new business models will be “crushed” in the marketplace. She believes healthcare AI spending will grow to $130 billion by 2025.16 U.S. hospitals are stepping up on behalf of their radiologists, with an estimated $2 billion in imaging AI spending projected by 2023.17  It's no wonder that imaging is emerging as the early leader in AI adoption. The four main characteristics that define successful AI adoption in business in according to a Deloitte study are front and center in imaging. These include a clear purpose; wise automation tasks; availability of clean, reliable data; and the opportunity to shift humans to higher social tasks.18 

5. When will medical imaging and AI be fully integrated

At the 2018 American Roentgen Ray Society meeting in April, a professional panel considered this very question.19 One panel member, Abdul Hamid Halabi, global lead for healthcare at NVIDIA, a gaming company involved in integrating AI into healthcare, said that big developments like a one-push button magnetic resonance scanner are still ways off. But he thinks that a development such as automatic scan protocol could emerge fairly soon. Another panel member, Dr. Eliot Siegel, professor of diagnostic radiology at the University of Maryland and co-creator of the world’s first filmless radiology department at the Baltimore Veteran's Affairs Medical Center, believes that radiologists will see major advances in imaging applications in the next five to 10 years. Siegel cited the speed at which specialty algorithm companies are developing competencies that are useful to the medical imaging sector. He likened the journey of AI in medical imaging to the current development of the driverless car: 90 percent of the algorithm is really easy, five percent is really hard and a fraction of one percent is really, really difficult.

References:

  1. AI in Medical Imaging: Opportunities And Hurdles. Med Device Online. https://www.meddeviceonline.com/doc/ai-in-medical-imaging-opportunities-and-hurdles-0001 . Last accessed August 6, 2018.
  2. Artificial Intelligence and Life in 2030: One Hundred Year Study on Artificial Intelligence: Report of the 2015-2016 Study Panel. Stanford University. http://ai100.stanford.edu/2016-report. Last accessed January 30, 2018.
  3. Artificial Intelligence for Medical Image Analysis -Companies-to-Action -2018. Frost & Sullivan. Last accessed October 23, 2018.
  4. Despite added AI costs, there are likely benefits for radiologists. HCB News https://www.dotmed.com/news/story/43875 . Last accessed August 6, 2018.
  5. Leveraging the algorithm: How imaging leaders can benefit from early AI adoption. The Advisory Board. https://www.advisory.com/research/imaging-performance-partnership/the-reading-room/2018/02/ai-radiology . Last accessed August 6, 2018.
  6. Survey: Majority of Imaging Leaders See Important Role for Machine Learning in Radiology. Healthcare informatics. https://www.healthcare-informatics.com/news-item/analytics/survey-majority-imaging-leaders-see-important-role-machine-learning-radiology . Last accessed August 6, 2018.
  7. Radiology organizations to work together to promote AI use. HealthData Management. https://www.healthdatamanagement.com/news/radiology-organizations-to-work-together-to-promote-ai-use?regconf=1 . Last accessed August 6, 2018.
  8. Radiology organizations to work together to promote AI use. HealthData Management. https://www.healthdatamanagement.com/news/wake-radiology-implements-platform-for-ai-algorithms . Last accessed August 6, 2018.
  9. AI in Medical Imaging: Opportunities And Hurdles. Med Device Online. https://www.meddeviceonline.com/doc/ai-in-medical-imaging-opportunities-and-hurdles-0001 . Last accessed August 6, 2018.
  10. Are you ready for a radiologist shortage? AuntMinnie.com. https://www.auntminnie.com/index.aspx?sec=sup&sub=aic&pag=dis&ItemID=120391. Last accessed August 6, 2018.
  11. The delicate balance between cost and quality in value-based healthcare. Becker’s Hospital CFO Report. https://www.beckershospitalreview.com/finance/the-delicate-balance-between-cost-and-quality-in-value-based-healthcare.html . Last accessed August 6, 2018.
  12. What is Value-based Healthcare? NEJM Catalyst. https://catalyst.nejm.org/what-is-value-based-healthcare/ . Last accessed August 6, 2018.
  13. AI in Medical Imaging: Opportunities And Hurdles. Med Device Online. https://www.meddeviceonline.com/doc/ai-in-medical-imaging-opportunities-and-hurdles-0001 . Last accessed August 6, 2018.
  14. Medical Imaging is Healthcare’s Artificial Intelligence Bellwether. HealthIT Analytics. https://healthitanalytics.com/news/medical-imaging-is-healthcares-artificial-intelligence-bellwether . Last accessed August 6, 2018.
  15. Analytics and AI in Radiology in the Era of Value-based Care. Link AHRA. https://link.ahra.org/2018/03/26/analytics-and-ai-in-radiology-in-the-era-of-value-based-care/ . Last accessed August 6, 2018.
  16. Future trends in healthcare. Healthcare Innovation. https://www.enterpriseinnovation.net/article/future-trends-healthcare-1342467221. Last accessed August 6, 2018.
  17. Hospitals to invest $2 billion every year in AI for diagnostic imaging. Healthcare Finance. https://www.healthcarefinancenews.com/news/hospitals-invest-2-billion-every-year-ai-diagnostic-imaging. Last accessed September 25, 2018.
  18. Leaders in cognitive and AI weigh in on what's working and what's next. Deloitte. https://www2.deloitte.com/content/dam/Deloitte/us/Documents/deloitte-analytics/us-da-2017-deloitte-state-of-cognitive-survey.pdf. Last accessed September 25, 2018.
  19. AI and the Future of Radiology. https://cmc-prod.s3-accelerate.amazonaws.com/attachments/3810adb0-d5ee-11e8-b975-a542a7d57098/15912-ARRS%20Lunch%20Symposium_v3b%5B1%5D%5B4%5D.pdf. Last accessed October 23,2018.