AI-enhanced breast imaging: Where are we and where are we heading?

Abstract

Significant advances in imaging analysis and the development of high-throughput methods that can extract and correlate multiple imaging parameters with different clinical outcomes have led to a new direction in medical research. Radiomics and artificial intelligence (AI) studies are rapidly evolving and have many potential applications in breast imaging, such as breast cancer risk prediction, lesion detection and classification, radiogenomics, and prediction of treatment response and clinical outcomes. AI has been applied to different breast imaging modalities, including mammography, ultrasound, and magnetic resonance imaging, in different clinical scenarios. The application of AI tools in breast imaging has an unprecedented opportunity to better derive clinical value from imaging data and reshape the way we care for our patients. The aim of this study is to review the current knowledge and future applications of AI-enhanced breast imaging in clinical practice.

Deep learning-based automatic segmentation for size and volumetric measurement of breast cancer on magnetic resonance imaging

“…Preoperative breast MRI, for its highest resolution and abundant information, becomes the most promising imaging modality for different AI applications, mainly for lesion detection and classification ( 12 , 29 ). Automatically detecting and classifying (limited to benign versus malignant) breast lesions on MRI are relatively well-established techniques ( 30 – 33 ).…”

“…Although the classification capability of our model is powerful and significant, future advances which will be considered through external validation in other institutions or with larger data sets will make it more persuasive for clinical application. Preoperative breast MRI, for its highest resolution and abundant information, becomes the most promising imaging modality for different AI applications, mainly for lesion detection and classification (12,29). Automatically detecting and classifying (limited to benign versus malignant) breast lesions on MRI are relatively well-established techniques (30-33).…”

Artificial Intelligence in Breast Ultrasound: The Emerging Future of Modern Medicine

“…The lack of a consistent strategy for segmentation (2D vs. 3D), feature extraction, and selection and categorization of significant radiomic data is a common limitation shared by all imaging modalities. Future studies with greater datasets will allow for subgroup analysis by patient group and tumor type [ 36 ].…”

A Machine Learning Ensemble Based on Radiomics to Predict BI-RADS Category and Reduce the Biopsy Rate of Ultrasound-Detected Suspicious Breast Masses

“…Radiomics, i.e., the measurement of a high number of quantitative features from images characterizing size, shape, image intensity, and texture of identified findings, has been extensively used to train multivariate machine learning algorithms to objectively characterize image findings and to predict diagnosis and prognosis of individual lesions or subjects. In breast cancer care, radiomics has been applied to a variety of medical image modalities for the aforementioned purposes, including mammography, digital breast tomosynthesis, ultrasound, magnetic resonance imaging, and positron-emission tomography combined with computed tomography [ 7 , 8 , 9 , 10 ], with good performances and with the advantage of high explainability, in particular when the radiomic predictors of the models can be compared and interpreted with reference to semantic predictors previously described in literature. In particular, many features of breast lesions on ultrasound images are known to be associated with higher or lower probability of malignancy of a given lesion, as Stavros et al [ 11 ] pointed out in their seminal paper focused on breast solid masses published more than 25 years ago.…”