AI governance: Ensuring ethical and responsible AI development

By Radiology AI Pillar members doctors Daniel Vergara, Kalpana Kanal, Nathan Cross and Jonathan Medverd

In Fall 2024, the Department of Radiology AI Pillar created a policy document that describes the process for managing medical imaging artificial intelligence (AI) tools, software, and algorithms applied to medical imaging in the practice of medicine across the UW healthcare environment. Dubbed Medical Imaging Artificial Intelligence Governance, this report describes the processes and key players involved to purchase, acquire, install, implement, monitor and decommission AI tools, software, and algorithms used with pixel-based medical imaging for clinical interpretation and diagnosis over their lifetime. The AI models and tools targeted by the AI Governance Policy are those which modify the images used in clinical evaluation of patient examinations and those that are used in computer aided detection (CAD) by the radiologist.

The AI governance policy is based on best practices and recommendations from sources like AAPM Task Group Report 273 (1), the AAPM CAD Subcommittee (2, 3, 4), IAEA Report “AI in Medical Physics,” (5), the ACR Recognized Center for Healthcare-AI (ARCH-AI) (6, 7) initiative and others (8, 9). It takes into consideration many guiding principles, some of which are listed below:

• make the right thing to do the easy thing to do
• be proactive (vs. reactive)
• implement formal and transparent operational processes
• embrace policies that are not overly restrictive but include clear consequences
• utilize a scalable approach, efficient, integrated and aligned with other governance

Our hope is that the AI Governance policy will provide useful guidance in the assessment and implementation process of new AI tools in the department of Radiology.

References 

1. Hadjiiski L, Cha K, Chan H-P, et al. Recommendations on best practices for AI and machine learning for computer-aided diagnosis in medical imaging. Med Phys 2023;50:e1-24. 

1. Vergara D, Armato SG, Hadjiiski L, et al. “Best Practices for Artificial Intelligence and Machine Learning for Computer-Aided Diagnosis in Medical Imaging.” Journal of the American College of Radiology, vol. 21, no. 2, 2024, pp. 341-343. 

1. Mahmood U, Shukla-Dave A, Chan HP, et al. Artificial intelligence in medicine: mitigating risks and maximizing benefits via quality assurance, quality control, and acceptance testing, BJR|Artificial Intelligence, 2024, 1(1), ubae003 https://doi.org/10.1093/bjrai/ubae003.  

1. R K Samala, K Drukker, A Shukla-Dave, et al. AI and machine learning in medical imaging: key points from development to translation, BJR|Artificial Intelligence, 2024, 1(1), ubae006 https://doi.org/10.1093/bjrai/ubae006. 

1. Artificial Intelligence in Medical Physics. IAEA, 2023, IAEA-TCS-83, Vienna. 

1. “American College of Radiology Launches First Medical Practice Artificial Intelligence Quality Assurance Program.” American College of Radiology, 26 June 2024, www.acr.org/Media-Center/ACR-News-Releases/2024/ACR-Launches-First-Medical-Practice-Artificial-Intelligence-Quality-Assurance-Program. 

1. “ACR Recognized Center for Healthcare-AI (ARCH-AI).” American College of Radiology, www.acrdsi.org/DSI-Services/ARCH-AI. Accessed 6 Nov. 2024. 

1. Daye, D., Wiggins, W.F., Lungren, M.P., Alkasab, T., Kottler, N., Allen, B., Roth, C.J., Bizzo, B.C., Durniak, K., Brink, J.A. and Larson, D.B., 2022. Implementation of clinical artificial intelligence in radiology: who decides and how?. Radiology, 305(3), pp.555-563.  

1. Omoumi, P., Ducarouge, A., Tournier, A., Harvey, H., Kahn, C.E., Louvet-de Verchère, F., Pinto Dos Santos, D., Kober, T. and Richiardi, J., 2021. To buy or not to buy—evaluating commercial AI solutions in radiology (the ECLAIR guidelines). European radiology, 31, pp.3786-3796.