Most people will have heard of genomics – the study of a person’s genes – or proteomics – the study of proteins – but what about the field of radiomics? And why is it important, especially in the context of breast cancer?  Simply stated, radiomics is a field of study that aims to extract quantitative features from medical images using mathematical algorithms. It has the potential to reveal characteristics of disease that may be difficult or impossible for the naked eye to discern. Radiomics consists of multiple steps: image acquisition; image segmentation; feature extraction; signature development; statistical analysis; and clinical applications.

Writing in BioMed Research International last year, Paolo Crivelli and colleagues from Italy reviewed the role of radiomics as a diagnostic tool in breast cancer. They focused on the ability of radiomics to predict malignancy, its response to neoadjuvant chemotherapy, prognostic factors, molecular subtypes of breast cancer, and the risk of recurrence. Twenty papers – which included the imaging modalities of mammography, ultrasound and MRI – were reviewed, and the authors demonstrated how the integration of radiomics with clinical, histological and genomic data could enable the provision of personalized treatment for breast cancer patients.

Earlier this year, Hui Li and co-authors from the University of Chicago published a paper in Radiology entitled, “Digital mammography in breast cancer: additive value of radiomics of breast parenchyma.” As seen in the diagram at left (© RSNA), they segmented both the tumour and a region of parenchymal tissue from the normal contralateral breast. What they found was that by combining quantitative radiomic data from tumours with data from the contralateral parenchyma, the diagnostic accuracy of breast cancer could be improved.

Next month’s European Journal of Radiology will showcase the findings of Chuqian Lei and colleagues from various institutions in China whose paper is entitled “Mammography-based radiomic analysis for predicting benign BI-RADS category 4 calcifications.” Using machine learning to select features and establish a radiomic signature, they built a nomogram that was able to discriminate between benign and malignant calcifications. In patients whose calcifications were negative on ultrasound but could be detected by mammography, the performance of their radiomic nomogram was judged to be “very strong.”

The Chicago team has also demonstrated that the combination of mammography radiomics with measurements of water-lipid-protein tissue composition has the potential to reduce unnecessary breast biopsies. In this era of artificial intelligence (AI) and big data, it would appear likely that radiomics will play an increasingly vital role in both the diagnosis and treatment follow-up of breast cancer patients.