Newsletter issue #5
Redefining what’s possible
_Spotlight
Overcoming AI obstacles in radiology
AI has already revolutionized radiology. Advanced machine learning models are redefining how we approach patient care—from improving diagnostic accuracy to personalizing treatment plans. However, the path to successful AI integration in radiology has challenges.
One of the biggest hurdles currently for developers is data quality and availability. Developing robust AI models requires large, diverse data sets, however, many institutions struggle with the volume and variety needed. This gap can slow model training, leaving researchers to rely on synthetic data. Yet, finding synthetic data that reflects real-world scenarios is difficult, leading to biased practices.
Another major challenge is the need for more transparency in AI models. Radiologists can use AI to support diagnosis, but they need to understand why a model suggests a certain outcome. The unknown nature of many AI algorithms limits trust and slows adoption. Though visual explanation tools, such as class activation maps (CAMs) and causality-based methods, are creating transparency to become more reliable and clinically relevant, they require further refinement.
Collaboration across institutions is key to building stronger AI models, but patient data privacy and security concerns can complicate this process. Federated learning—a method that allows institutions to train models together without sharing sensitive data—is a potential solution, but technical and operational challenges limit adoption.
Mayo Clinic Platform provides secure, scalable access to high-quality, de-identified clinical data sets critical for training accurate AI models, working to overcome barriers to adoption, through their federated data network, Mayo Clinic Platform_Connect. By keeping data within each partner’s local environment, Mayo Clinic Platform_Connect upholds strict privacy standards while facilitating collaboration. Integrated multimodal imaging with longitudinal clinical data allows for more valuable connections and deeper insights. With these innovative tools, curated data sets, and a secure environment, Mayo Clinic Platform gives developers the resources to create new AI solutions for radiology.
To learn more about how Mayo Clinic Platform supports companies using imaging data, meet with us at RSNA in Chicago, Dec. 1-5!
_News
Mayo Clinic Researchers To Present on Artificial Intelligence in Medical Imaging
Mayo Clinic researchers Dr. Andrew Norgan and Dr. Panagiotis Korfiatis discuss how AI is changing medical imaging, including the latest uses of AI in radiology and pathology, what’s on the horizon, and how new tech like large language models is helping make diagnoses more accurate and efficient.
AI Could Be a Game Changer, but Healthcare Needs To Be ‘Exceedingly Careful’
Artificial intelligence tools could help solve workforce challenges. Implementation, however, can be difficult, pushing organizations to consider less risky administrative and back-office tasks first.
_Featured this month
BLOG
Challenging the status quo in technology and medicine can take a heavy toll, but history has demonstrated that it often yields measurable benefits for both clinicians and patients.
BLOG
Mayo Clinic Platform_Connect is transforming how patient data is used to generate innovative diagnostic tools and treatment options.
_In case you missed it
