Author: Wilfred R Furtado, Gary Y. Ge, James Lee, Jie Zhang 👨🔬
Affiliation: University of Kentucky 🌍
Purpose: Despite advancements in Artificial Intelligence (AI) and its growing role in clinical practices like radiology, formal AI education remains limited in medical training. This gap contributes to inefficiencies and reduced confidence among practitioners when incorporating AI into their daily workflows. To address this need, we introduced an 8-hour hands-on AI Education for first year radiology residents in 2023 and 2024. This study summarizes the structure, content, and initial outcomes of the course.
Methods: The course covered foundational AI concepts, including radiomics, convolutional neural networks, model training, validation, and performance evaluation. The program was divided into three sessions: convolution and image processing (1 hour), radiomics and machine learning (2 hours), and deep learning (5 hours). Each session combined short lectures, hands-on coding exercises, and clinical application discussions. Residents explored the impact of AI on clinical practice, such as CT kernel applications and its influence on image quality and decision-making. The second iteration of the course in Fall 2024 followed a similar format, emphasizing interactive learning using Python coding and practical implementation in Google Colab. Residents' understanding and interest were evaluated through post-training assessments.
Results: Post-training assessments demonstrated significant improvements in residents' confidence with AI concepts and their understanding of AI applications in radiology. Participants particularly valued the hands-on coding exercises and expressed increased interest in further exploring AI tools. The course also established a framework for systematic AI education, including structured modules with learning objectives, illustrative examples, and review questions tailored for clinical workflows.
Conclusion: The hands-on AI Education effectively introduces radiology residents to foundational AI concepts, bridging a critical gap in training. This initiative lays the groundwork for integrating AI education into residency programs, equipping residents for confident and effective use of AI tools in clinical practice.