Radiomics-based analysis of contrast-enhanced and digital mammography for breast cancer classsification

Abstract

Contrast-Enhanced Spectral Mammography (CESM) is an advanced imaging modality that enhances breast cancer detection by combining conventional digital mammography (DM) with CESM obtained through intravenous contrast administration. This dual approach provides both morphological and functional information, improving lesion visibility, particularly in patients with dense breast tissue. Radiomics, a rapidly evolving field in medical imaging, allows the extraction of high-dimensional quantitative features from medical images, capturing information about tumour phenotype, texture, and heterogeneity that may not be visually apparent. This thesis investigates the application of radiomics analysis to CESM images with the aim of improving breast cancer classification. A key focus is on comparing the diagnostic performance of radiomics features derived separately from CESM and DM images, as well as evaluating the added value of combining both sets of features. Radiomics features are extracted and analysed using statistical and traditional machine learning techniques to assess their effectiveness in distinguishing between benign and malignant lesions. Furthermore, the study explores the development of predictive models based on these features and identifies the most relevant biomarkers for tumour classification. By systematically evaluating radiomics features derived from CESM, DM, and their combination, this research aims to determine the most effective imaging strategy for accurate breast cancer classification. The findings may support more informed clinical decision-making and contribute to the advancement of personalized diagnostic approaches in breast cancer care. Results show that models based on CESM images consistently outperformed those based on DM and combined data, confirming the hypothesis that contrast-enhanced imaging yields more informative radiomic features for tumour classification.