Artificial Intelligence for Breast MRI in 2008-2018: A Systematic Mapping Review

Area of Science:

• Radiology and Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Breast MRI is a crucial imaging modality for breast cancer detection and management.
• Artificial intelligence (AI) offers potential advancements in analyzing complex breast MRI data.
• The application of AI in breast MRI has seen significant growth over the last decade.

Purpose of the Study:

• To conduct a comprehensive literature review on the applications of artificial intelligence (AI) in breast MRI over the past ten years.
• To identify and analyze the trends, methods, and clinical aims of AI in breast MRI research.

Main Methods:

• A systematic literature search was performed in June 2018 to identify relevant studies.
• Data extracted included study design, AI methods, clinical aims, and diagnostic performance.
• Sixty-seven studies published within the last decade were analyzed.

Main Results:

• The majority of studies (87%) were retrospective, with 36% published in radiology and imaging journals.
• Contrast-enhanced sequences were predominantly used (75%).
• Key applications included breast lesion classification (54%), image processing (21%), prognostic imaging (13%), and response to neoadjuvant therapy (12%). Artificial neural networks and support vector machines were common algorithms. Supervised learning achieved high AUCs for lesion characterization (median 0.87) and prognostic imaging (median 0.80).

Conclusions:

• Interest in AI for breast MRI is increasing globally.
• Current AI applications demonstrate promising diagnostic performance but require further development for widespread clinical adoption.
• Future research should focus on refining AI algorithms and validating their performance in clinical settings.