Machine learning for multi-parametric breast MRI: radiomics-based approaches for lesion classification.

Area of Science:

• Artificial Intelligence in Medicine
• Radiomics and Medical Imaging Analysis
• Oncology and Breast Cancer Diagnostics

Background:

• Machine learning (ML) is crucial for analyzing complex medical image data.
• Radiomics extracts numerous features from medical images, linking tumor phenotype to genomic pathways.
• Multi-parametric breast MRI is vital for dense breast imaging and high-risk patient screening.

Purpose of the Study:

• To review the application of ML techniques in multi-parametric breast MRI.
• To focus on ML for tumor classification and differentiation of molecular subtypes in breast cancer.
• To provide an overview of current ML models, their advantages, drawbacks, and performance.

Main Methods:

• Review of recent literature on ML in multi-parametric breast MRI.
• Analysis of various ML models and approaches used for breast cancer classification.
• Evaluation of techniques for differentiating molecular subtypes based on imaging features.

Main Results:

• ML techniques are rapidly expanding in breast MRI, improving diagnostic and prognostic capabilities.
• Radiomics-derived features combined with ML show potential for detailed tumor characterization.
• Different ML models exhibit varying performance in classifying breast tumors and subtypes.

Conclusions:

• ML applied to multi-parametric breast MRI offers significant advancements in breast cancer diagnosis.
• Further research is needed to optimize ML models for robust tumor classification and subtype differentiation.
• Understanding the strengths and weaknesses of different ML approaches is key for clinical translation.