Go to main navigation Navigation menu Skip navigation Home page Search

HOI research | Holistic approach to AI in radiology to overcome implementation hurdles

Can a comprehensive strategy make AI work in radiology? A new study in Insights into Imaging shows how addressing technology, workflow, and organization challenges can lead to successful AI integration.

(This image was generated using AI.)

AI in radiology faces multifaceted challenges

AI holds great promise for transforming radiology, improving diagnostic accuracy and efficiency. However, integrating AI into clinical practice is complex, facing obstacles at multiple levels. At the technology level, AI applications offer narrow functions with no standard user interface, resulting in issues around scalability and integration. At the workflow level, AI applications generate results with limited modifiability and correspondence with the organization's specific patient population. At the people and organization level, there are varying expectations and limited experience among clinicians regarding AI. This study, conducted over three years at a major university medical center in the Netherlands, shows how these multifaceted challenges can be addressed through a holistic implementation approach.

How a holistic approach can help

By conducting extensive observations, interviews, and data analysis, the researchers sought to demonstrate how a comprehensive strategy could overcome these challenges around integrating AI in radiology. The goal was to provide actionable insights that can be transferred to other healthcare organizations and help them successfully integrate AI into their clinical practices.

"Integrating AI into the daily operations of a complex healthcare system has proven to be enormously challenging across technology, workflow, and organizational levels. This case study illustrates the effectiveness of a holistic approach to AI implementation in addressing these challenges and achieving sustainable value creation." — Dr. Bomi Kim, Assistant Professor, House of Innovation.

Key research findings

  • Implementing a vendor-neutral AI platform helps streamline AI integration by automatically routing data and centrally managing legal and technical procedures.
  • Allowing radiologists to modify AI-generated results increases their control and trust in the technology, encouraging its use.
  • Establishing a dedicated team, like the Image Processing Group, ensures consistent use of AI and builds expertise among radiographers and technical physicians.

Towards widespread AI adoption in healthcare

This study underscores the importance of a holistic approach in successfully implementing AI in radiology. By addressing technological, workflow, and organizational challenges comprehensively, healthcare organizations can create sustainable value from AI. Future research should explore such approaches' financial viability and scalability to facilitate broader AI adoption across various medical fields.

Meet the researchers:

  • Dr. Bomi Kim: House of Innovation (Department of Entrepreneurship, Innovation and Technology), Stockholm School of Economics
  • Dr. Stephan Romeijn: Radiology, Leiden University Medical Center
  • Dr. Mark van Buchem: Radiology, Leiden University Medical Center
  • Dr. Mohammad Hosein Rezazade Mehrizi: KIN Center for Digital Innovation, Vrije Universiteit Amsterdam
  • Dr. Willem Grootjans: Radiology, Leiden University Medical Center
House of Innovation Health Digitalization Innovation Article Journal Publication Research