[Artificial intelligence in breast imaging : Hopes and challenges].

Abstract

Clinical/methodical issue: Artificial intelligence (AI) is being increasingly integrated into clinical practice. However, the specific benefits are still unclear to many users.

Standard radiological methods: In principle, AI applications are available for all imaging modalities, with a particular focus on mammography in breast diagnostics.

Methodical innovations: AI promises to filter examinations into negative and clearly positive findings, and thereby reduces part of the radiological workload. Other applications are not yet as widely established.

Performance: AI methods for mammography, and to a lesser extent tomosynthesis, have already reached the diagnostic quality of radiologists.

Achievements: Except for second-opinion applications/triage in mammography, most methods are still under development.

Practical recommendations: Currently, most AI applications must be critically evaluated by potential users regarding their maturity and practical benefits.