By Mark Watts
I’m always on the lookout for opportunities to help you elevate your expertise and connect with the brightest minds in our field. Mark your calendars because SIIM 2025 in Portland, Oregon, from June 11-14, is shaping up to be an unmissable event! Beyond the beautiful scenery and fantastic food that Portland offers, my hometown. SIIM is bringing together a powerhouse of presenters ready to share their invaluable insights.
Let me give you a sneak peek at some of the luminaries you’ll have the chance to learn from:
- Paul Nagy: A true visionary in medical imaging informatics, Dr. Nagy consistently pushes the boundaries of what’s possible. His sessions are always thought-provoking, offering a glimpse into the future of our field. Why attend his presentation? Expect to gain a deeper understanding of cutting-edge technologies and strategic approaches that can transform your practice.
- Dawn Cram: With her extensive experience in clinical informatics, she brings a practical and insightful perspective to the challenges and opportunities we face daily. Why attend her presentation? Look forward to actionable strategies and best practices that you can implement immediately to improve workflow and efficiency.
- Woojin Kim: Dr. Kim is a leading expert in artificial intelligence and its application in medical imaging. His presentations demystify complex AI concepts and showcase their real-world impact. Why attend his presentation? Prepare to be inspired by the potential of AI to revolutionize diagnostics and patient care, and gain clarity on how to navigate this rapidly evolving landscape.
- Teri Sippel Schmidt: Sippel Schmidt’s expertise in imaging informatics leadership and education is unparalleled. She has a knack for empowering individuals and teams to excel. Why attend her presentation? You’ll gain valuable insights into effective leadership strategies and practical approaches to fostering growth and innovation within your department.
- Nina Kottler: Dr. Kottler is a renowned radiologist and researcher with a deep understanding of the clinical applications of imaging informatics. Her presentations bridge the gap between technology and patient care. Why attend her presentation? Expect to learn about the latest clinical advancements and how informatics plays a crucial role in delivering optimal patient outcomes.
- Judy Gichoya: Dr. Gichoya is a passionate advocate for addressing bias and promoting equity in AI and medical imaging. Her work is critical in ensuring that technological advancements benefit all patient populations. Why attend her presentation? Be prepared for an eye-opening discussion on the ethical considerations of AI and how we can build more inclusive and equitable imaging systems.
So, why should YOU choose their presentations?
Because these aren’t just lectures; they are opportunities to learn from the best, engage in stimulating discussions, and gain knowledge that can directly impact your work and your patients’ lives. You’ll walk away with fresh perspectives, practical tools, and a renewed sense of purpose in your vital role in the healthcare ecosystem.
Don’t miss out on this incredible chance to learn, connect, and grow at SIIM 2025 in Portland. The topic I would recommend concentrating on is the importance of Pre-Determined Change Control Plans by the FDA in Medical Imaging Artificial Intelligence.
PCCPs for Imaging AI by the FDA: A Shift
Following our discussion on the importance of Pre-Determined Change Control Plans (PCCPs), let’s now focus on how the FDA’s approach to PCCPs specifically for imaging artificial intelligence (AI) differs significantly from current practices for traditional medical devices and even earlier AI implementations. This shift reflects the unique dynamic and adaptive nature of AI algorithms.
Currently, the traditional FDA pathway for medical devices, including software, often operates on a “lock-and-key” principle. Once a device receives clearance or approval, significant changes, especially those affecting performance or intended use, typically require a new premarket submission (e.g., a 510(k) supplement or a new De Novo application). This model can be cumbersome and slow the iterative improvement that is inherent to AI development.
Here’s how FDA’s PCCP framework for imaging AI marks a departure from this traditional approach:
1. Embracing Adaptability and Continuous Improvement
- Current Practice: Traditional devices and even early “locked” AI algorithms are largely static after-market authorization. Updates often necessitate a new review cycle.
- PCCP for Imaging AI: The FDA recognizes that AI algorithms, particularly machine learning-based ones, are designed to learn and improve over time with new data. PCCPs provide a mechanism to prospectively outline anticipated modifications to the AI model (e.g., retraining with new datasets, adjusting model parameters) within pre-defined boundaries, without requiring a new submission for each change. This fosters a life cycle approach to regulation.
2. Prospective Review of Planned Changes
- Current Practice: Changes are typically reviewed after they are developed and implemented (or planned for implementation), often triggering a regulatory hurdle.
- PCCP for Imaging AI: Manufacturers are now encouraged to submit a detailed PCCP as part of their initial premarket submission. This plan outlines the types of modifications they foresee, the protocols for evaluating these changes (including validation and testing metrics), and the criteria for determining if a modification falls within the pre-approved scope. The FDA reviews and authorizes this plan upfront.
3. Risk-Based Approach to Change Management
- Current Practice: All significant changes might trigger a similar level of regulatory scrutiny, regardless of the potential risk.
- PCCP for Imaging AI: The PCCP framework emphasizes a risk-based approach. Manufacturers need to clearly articulate the potential impact of planned modifications on the device’s safety and effectiveness. The level of detail and rigor required in the PCCP will likely be proportional to the potential risk associated with the anticipated changes.
4. Focus on Transparency and Labeling
- Current Practice: Device labeling typically describes the device’s functionality at the time of market authorization.
- PCCP for Imaging AI: The FDA emphasizes the need for clear labeling that informs users about the AI’s adaptive nature and the existence of an authorized PCCP. This ensures users are aware that the device’s performance or outputs may evolve over time through software updates.
5. Emphasis on Performance Monitoring and Validation Protocols
- Current Practice: Post-market surveillance often relies on adverse event reporting and periodic reviews.
- PCCP for Imaging AI: PCCPs require manufacturers to define robust protocols for monitoring the AI’s performance in the real world and validating the effectiveness and safety of any implemented changes according to the pre-approved plan. This signifies a more proactive approach to ensuring ongoing safety and efficacy.
In essence, PCCPs for imaging AI represent a shift from a static, reactive regulatory model to a more dynamic, proactive, and life cycle-oriented approach. This framework aims to balance the need for innovation in this rapidly evolving field with the FDA’s mandate to ensure the safety and effectiveness of medical devices. It requires manufacturers to think ahead, plan for change, and demonstrate a robust understanding of how their AI algorithms will be managed and monitored throughout their life cycle.
This evolution in regulatory strategy is crucial for realizing the full potential of AI in medical imaging while maintaining the highest standards of patient care. As we learn more at SIIM 2025, understanding these shifts will be paramount for all stakeholders in the imaging AI ecosystem.
Mark Watts is an experienced imaging professional who founded an AI company called Zenlike.ai.
