J Am Coll Radiol. 2019 Jul;16(7):961-963. doi: 10.1016/j.jacr.2019.04.023. Epub 2019 May 16.
Allen B, Agarwal S, Kalpathy-Cramer J, Dreyer K.
Whether it is adjusting parameters for MR pulse sequences, specifying parameters to optimize radiation exposure on CT scanners, or programming complex treatment planning protocols in radiation oncology, radiology professionals use sophisticated computer software platforms in our practices daily to take better care of our patients. Yet because most of us are not data scientists or software engineers, up until now, most of us feel at least a little intimidated by artificial intelligence (AI) and how we will be able to use it. We tend to believe that it takes an informatics background and training in complex computer programming languages such as Python (Python Software Foundation, Beaverton, Oregon) [1] to understand how AI algorithms are developed, how we can apply them in daily clinical practice, and how we can contribute. This needs to change. To ensure AI reaches its full potential in helping us take better care of our patients, all radiology professionals need to be able to understand how to use AI in our practices. We need to know about the various types of AI algorithms, how they work, their limitations, and how to evaluate them for clinical use. Radiology professionals also need to play an active role in bringing AI tools into routine clinical care. Whether it is participating in AI use case development, providing cases for algorithm training, participating in algorithm validation, or actually building AI models, our professions should be at the center of ensuring that AI deployed for clinical use is safe and effective and helps us solve the problems important to our patients and health systems.
Read Full Article Here: https://doi.org/10.1016/j.jacr.2019.04.023