Radiomics: A Primer for Breast Radiologists

Area of Science:

• Breast Imaging
• Radiomics
• Artificial Intelligence in Medicine

Background:

• Computer-aided detection (CAD) in mammography has a long history but faced limitations like high false positives.
• Recent advancements in artificial intelligence (AI), machine learning, and deep learning are revolutionizing radiomics in breast imaging.
• Radiomics applications span various modalities including mammography, MRI, ultrasound, and PET-CT.

Purpose of the Study:

• To review the evolution and current state of radiomics in breast imaging.
• To highlight diverse radiomics outcomes such as CAD, therapy response prediction, and lesion classification.
• To emphasize the growing importance of radiogenomics and its correlation with genetic data.

Main Methods:

• Review of extensive radiomics research across multiple breast imaging modalities.
• Exploration of radiogenomics, linking imaging features with genetic information.
• Analysis of challenges including image normalization, data set limitations, and external validation needs.

Main Results:

• AI-driven radiomics offers significant improvements over conventional CAD for breast imaging.
• Radiomics shows promise in predicting treatment response, classifying lesions, and assessing survival.
• Despite challenges, some radiomics tools are transitioning into clinical practice.

Conclusions:

• Radiomics, powered by AI, presents numerous opportunities for advancing breast cancer diagnosis and management.
• Radiologists need a comprehensive understanding of radiomics' scope, applications, and limitations.
• The integration of radiomics-based tools into daily practice is imminent, requiring radiologist preparedness.