Artificial intelligence is now ready to be implemented in our daily practice. Studies are showing that the currently available AI algorithm performances, especially for mammography, are at the level of radiologists. We are currently experimenting many ways to use AI so that it may help radiologists by reducing interpretation workload or time, while at the same time providing more accurate interpretation results to our patients. AI for mammography interpretation can be used for reader assistance or for triage, even considered to replace a second reader/arbitrator in double reading considering its high-level performances. While this prospect is quite exciting, there are still potential hurdles to overcome before considering widespread AI application, that will be discussed in this session.
Learning Objectives:
To summarize various features that can affect AI results when using AI for mammography
To discuss different aspects in the real-world to study trials that should be considered prior to AI implementation
To discuss future issues that needs to be solved for widespread AI implementation in breast radiology
