Integration of advanced ultrasound techniques, radiomics, and artificial intelligence for improved diagnosis of cystic breast lesions.
Authors
Affiliations (2)
Affiliations (2)
- Department of Diagnostic, Molecular and Interventional Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, USA.
- Department of Diagnostic, Molecular and Interventional Radiology, Icahn School of Medicine at Mount Sinai, New York, NY, USA. Electronic address: [email protected].
Abstract
Cystic breast lesions are commonly encountered on breast ultrasound, encompassing a spectrum from simple benign cysts to complex mixed solid and cystic lesions with malignant potential. Accurate sonographic characterization using ACR BI-RADS v2025 criteria is essential to guide clinical management, avoid unnecessary biopsies, and reduce patient anxiety. This review examines the sonographic features of cystic breast lesions across the pathological spectrum, including simple cysts, fat necrosis, abscesses, papillomas, and malignant subtypes such as medullary, papillary, and mucinous carcinomas. Key ultrasound techniques - including harmonic imaging, Doppler evaluation, and elastography - are discussed in the context of optimizing lesion characterization and differentiating benign from malignant cystic lesions. Despite these advances, image interpretation remains operator-dependent and challenging, particularly when malignancies mimic benign cystic features. Ultrasonomics emerged as a promising quantitative approach, extracting high-dimensional imaging features related to lesion shape, texture, and intensity to enable objective, reproducible assessment of lesion heterogeneity beyond visual interpretation. When combined with machine learning algorithms, ultrasonomics enhances lesion classification and diagnostic consistency. Building upon this framework, AI-based decision support systems analyze multiple sonographic parameters simultaneously, reducing interobserver variability and helping stratify malignancy risk in indeterminate or mixed solid and cystic lesions demonstrating suspicious features such as irregular walls, mural nodules, or internal vascularity. Clinical studies demonstrate AI can reduce unnecessary biopsies without compromising cancer detection rates. The integration of advanced ultrasound techniques, ultrasonomics, and AI-based decision support represents a meaningful advancement in the objective evaluation of cystic breast lesions, supporting more accurate risk stratification and optimized patient management.