Application of Deep Learning in Breast Cancer Imaging

Area of Science:

• Radiology
• Artificial Intelligence
• Medical Imaging

Background:

• Breast imaging is crucial for early cancer detection, treatment monitoring, and response assessment.
• Digitalization of imaging modalities like mammography, ultrasound, and MRI enables AI integration.
• Deep learning (DL) is a rapidly advancing subset of AI with significant potential in medical diagnostics.

Purpose of the Study:

• To provide a comprehensive overview of current deep learning research in breast cancer imaging.
• To highlight the applications of DL across various breast imaging techniques and diagnostic tasks.
• To identify the current limitations and future directions for DL in clinical breast care.

Main Methods:

• Review of existing literature on deep learning applications in breast cancer imaging.
• Analysis of DL performance in tasks including lesion classification, segmentation, image reconstruction, and risk prediction.
• Evaluation of DL's role in different imaging modalities: mammography, digital breast tomosynthesis, ultrasound, MRI, and nuclear medicine.

Main Results:

• Deep learning algorithms demonstrate comparable or superior performance to radiologists in several breast imaging tasks.
• DL is applied to lesion classification, segmentation, image generation, risk prediction, and therapy response assessment.
• Research on DL in nuclear medicine imaging is limited, requiring further investigation.

Conclusions:

• Deep learning holds significant potential to enhance breast cancer detection and management.
• Large-scale clinical trials are essential to validate DL's added value, particularly in ultrasound and MRI.
• Addressing legal, ethical, and regulatory aspects is critical for the widespread adoption of DL in breast imaging practice.