Conference on Machine Intelligence in Medical Imaging

Key factors that hinder the clinical applicability of AI technologies in radiology and radiation oncology include the nature and quality of the data used to train AI models, the recognition of the clinical limitations of these models, regulatory considerations specific to AI in medicine, and biases - both from clinicians and the automation itself. Furthermore, concerns about potential software errors and the ongoing maintenance of AI systems highlight the need for robust workflow monitoring and the tracking of critical performance parameters such as accuracy and reliability.

The medical physicist plays a crucial role in navigating these challenges, ensuring patient safety by verifying that AI is applied appropriately within the clinical context and produces results that meet clinical standards. Additionally, it is essential for medical physicists to set realistic expectations about what AI can achieve in clinical practice, balancing innovation with practicality. Medical physicists’ responsibilities will expand, necessitating greater collaboration with AI developers and other stakeholders to successfully navigate this evolving landscape.